专利摘要:
A wind farm comprising a plurality of wind turbines including a running wind turbine and a microphone located close to the wind farm. The wind farm also includes a computer device coupled to the microphone. The computing device includes a processor and a memory device coupled to the processor. The computing device is configured to produce (604) a noise recording by recording a plurality of wind farm sound pressure measurements using the microphone and to calculate (606) values of a plurality of acoustic characteristics associated with the noise. noise recording. The computing device is also configured to determine (612), using the computing device, a relationship between the calculated values for the plurality of acoustic characteristics and the modeled acoustic characteristic values present in a probabilistic acoustic model of the wind farm. The computing device is further designed to distinguish (614) a first contribution to noise recording by the running wind turbine from a second contribution to noise recording by sources other than wind turbines, according to the predetermined relationship.
公开号:FR3024502A1
申请号:FR1557381
申请日:2015-07-31
公开日:2016-02-05
发明作者:Andreas Herrig；Jean Benoit Philippe Petit；Sara Simonne Louisa Delport；Nathan Charles Schneider
申请人:General Electric Co；
IPC主号:

专利说明:

[0001] The invention relates generally to wind turbine generators and, more particularly, to a system and method for operating wind farms with improved acoustical measurements. SUMMARY OF THE INVENTION Most prior art wind turbines include a rotor with multiple blades. The rotor is sometimes mounted on a housing, or nacelle, placed (e) at the top of a support, for example a tubular mast. At least some prior art industrial wind turbines, i.e., wind turbines designed to supply electricity to a distribution network, have rotor blades of a predetermined shape and size. The rotor blades transform the mechanical energy of the wind into induced blade lift forces which further induce a mechanical torque of rotation resulting in one or more wind turbines, which then produce electricity. A plurality of wind turbines in a localized geographic area are commonly referred to as a wind farm.
[0002] During the operation of these wind farms according to the prior art, the rotation of the rotor blades under the action of the air creates aerodynamic acoustic emissions, namely noise. Therefore, at least some of these prior art wind farms will be provided with noise receiving devices in the vicinity of the wind farms to measure the overall noise level. At least some of these measured acoustic noises have a decibel level (dB) that may approach local regulatory thresholds. In order to meet the thresholds, it may be necessary to put at least some of the wind turbines in low noise operation (FBR) mode for a certain time. In this way, the reduction of noise is obtained to the detriment of the annual production of energy. Therefore, it is necessary to use RBFs that very effectively reduce noise levels for as short a time as possible and use them in a small number of turbines in the wind farm to achieve the desired acoustic levels. Prior art methods of achieving regulatory compliance include the use of far-field sound propagation models, based on certain site parameters, for example turbine-receiver distances, ground absorption, shear wind and thermal gradients associated with a model of turbine noise. However, too cautious parameters may be chosen and needlessly reduced production. Another prior art method involves a direct measurement of the acoustic environment in the vicinity of the wind farm and the use of a servo control to regulate the wind turbines to FBR mode or to exit the mode. FBR while taking into account time-related changes in turbine noise levels, eg air density, blade fouling condition, wind shear and propagation characteristics. However, this measurement-based control lacks the means to distinguish whether the measured noise actually originates from the wind turbines or is contaminated and possibly dominated by ambient noise. The latter situation usually results in a biased estimate of the sound pressure level (NPA), and the servo means then tend to seek to reduce the noise of wind turbines via the NPA even if the noise levels of wind turbines are well below established parameters. Therefore, servo settings (as well as the processing of certification measures) must include extensive acoustic measurements to identify and disregard sound recordings contaminated by foreign sound contributions, such as automobiles, aircraft and birds. These activities usually involve lengthy listening checks and manual removal of contaminated segments. As a result, the most pronounced sound aberrations can easily be identified and neglected by examining the difference in dB levels between the measured noise and the peak of the expected wind turbine noise curve, or the complete curve if the speed ratio of the wind / speed of the wind turbine is known. However, variations in the NPA over long distances due to propagation effects introduce high uncertainty if only this type of filtering is used, which reduces the usefulness of the data. According to a first aspect, there is provided a computer method for operating a wind farm. The method uses a computing device having at least one processor coupled to a memory device and the wind farm has at least one wind turbine running. The method includes recording a plurality of sound pressure measurements of the wind farm, which creates a noise record. The method also includes calculating values for a plurality of acoustic features associated with noise recording. The method further includes determining a relationship between the calculated values for the plurality of acoustic characteristics and the modeled acoustic characteristic values present in a probabilistic acoustic model of the wind farm. The method also distinguishes, according to the determined relationship, between a first contribution to the recording of the noise emanating from the at least one wind turbine running relative to a second contribution to noise recording from other sources. than wind turbines.
[0003] In another aspect, a wind farm is proposed. The wind farm has a plurality of wind turbines including a wind turbine running. The wind farm also has a microphone located near the wind farm and a computer device coupled to the microphone. The computing device includes a processor and a memory device coupled to the processor. The computing device is designed to create a noise record by recording a plurality of acoustic pressure measurements of the wind farm using the microphone. The computing device is also adapted to calculate values for a plurality of acoustic features associated with noise recording. The computing device is further adapted to determine, using the computing device, a relationship between the calculated values for the plurality of acoustic characteristics derived from noise recording with modeled acoustic characteristic values present in a probabilistic acoustic model. of the wind farm. The computing device is also designed to distinguish, from the predetermined relationship, a first contribution to noise recording from the running wind turbine with a second contribution to recording noise from sources other than wind turbines. In yet another aspect, one or more computer-readable storage media containing computer executable instructions is / are provided. When executed by the processor (s), the computer-executable instructions cause the processor (s) to create a noise record by recording a plurality of sound pressure measurements of the wind farm. The computer executable instructions also cause the processor (s) to calculate values for a plurality of acoustic characteristics associated with noise recording. The computer-executable instructions further cause the processor (s) to determine a relationship between the calculated values for the plurality of acoustic characteristics and the acoustic characteristic values present in the probabilistic acoustic model of the wind farm. The computer-executable instructions also cause the processor (s) to distinguish, from the predetermined relationship, a first contribution to the recording of noise emanating from a running turbine with a second contribution to the recording. noise from sources other than wind turbines.
[0004] The invention will be better understood on studying the detailed description of an embodiment taken by way of nonlimiting example and illustrated by the drawings in which: FIG. 1 is a schematic diagram of an example of computer device; Figure 2 is a block diagram of a portion of an exemplary wind farm control system that may include the computing device shown in Figure 1; 3 is a schematic view of an example of a wind turbine that can be used in a wind farm, both of which can be controlled and controlled using the wind farm control system shown in FIG. 2; - Figure 4 is a schematic view of an example of a wind farm which may include the wind turbine shown in Figure 3; Figure 5 is a schematic view of an exemplary wind farm noise control and regulation system operable with the wind farm control system shown in Figure 2; FIG. 6 is an example of a view of variations in the level of noise likely to be associated with the wind farm shown in FIG. 4; Figure 7 is an exemplary view of a probabilistic acoustic model for determining acceptable recordings and unacceptable recordings using the wind farm noise control and regulation system shown in Figure 5; Figure 8 is a flowchart of an exemplary method for operating the wind farm shown in Figure 3 using the wind farm noise control and regulation system shown in Figure 5; FIG. 9 is an example of a graphical view of noise data estimates as a function of a mean asymmetry and a mean fluctuation using a wind farm noise control and regulation system shown on FIG. Figure 5, having a different architecture; FIG. 10 is an example of a graphical view of noise data estimates as a function of standard deviations of flux and of a mean fluctuation using a wind farm noise control and regulation system represented in Figure 5, having another architecture; Fig. 11 is an example of a graphical view of comparisons of different methods for determining acceptable recordings and unacceptable recordings using the wind farm noise control and regulation system architecture shown in Fig. 5; and Fig. 12 is an exemplary configuration of a database in the computing device shown in Fig. 1, together with other associated computer components, which can be used to determine acceptable records and unacceptable records as described. right here. Unless otherwise indicated, the drawings presented herein are intended to illustrate aspects of embodiments of the present invention. These aspects are believed to be useful in all kinds of systems comprising one or more embodiments of the present invention. In this way, the drawings are not intended to cover all the conventional aspects which those of ordinary skill in the art know are necessary for the implementation of the embodiments presented herein. In the description and the claims which follow, reference will be made to a number of terms which are to be defined as having the following meanings. Articles indefinite and defined cover references to the plural, unless the context clearly requires the opposite. The adjective "possible" or the adverb "possibly" means that the fact or circumstance to be described may or may not occur and that the description covers cases where the fact occurs and cases where it does not occur. does not occur.
[0005] Approximate terms used herein throughout the specification and claims may be used to modify any quantitative representation that could possibly vary without causing a change in the basic function to which it corresponds. In this way, a value modified by one or more terms such as "approximately", "approximately" and "substantially" should not be limited to the precise value indicated. In at least some cases, the vocabulary expressing an approximation may correspond to the precision of an instrument used to measure the value. Here and in the entirety of the description and claims, range boundaries may be combined and / or interchanged, these ranges being identified and covering any secondary intervals they include, unless the context or vocabulary used 'indicates the opposite. For the purpose of the present description, the terms "processor" and "computer" and related terms, for example "processing device", "computing device" and "PLC" are not limited merely to integrated circuits known as computer in the art but more broadly refer to a microcontroller, a microcomputer, a programmable logic controller (PLC), an integrated circuit for specific application and other programmable circuits, and these terms are used interchangeably herein. In the embodiments described herein, a memory may include, but is not limited to, a computer-readable medium such as random access memory (RAM) and a computer-readable remanent medium such as flash memory. According to other possibilities, it is also possible to use a floppy disk, a CD-ROM, a magneto-optical disk (MOD) and / or a digital versatile disk (DVD). On the other hand, in the embodiments described herein, additional input channels may, but in no way limit, consist of computer peripherals associated with an operator input interface such as a mouse and a keyboard. Alternatively, other computer peripherals may be used which may include, for example, but in a non-limiting manner, a scanner. Further, in the exemplary embodiment, additional output channels may include, but are not limited to, an interface such as an operator input screen. Moreover, for the purposes of this description, the terms "software" and "firmware" are interchangeable and refer to any computer program stored in memory for execution by personal computers, workstations, clients and servers. For the purposes of the present description, the term "computer-readable remanent media" is intended to mean any hardware computing device implemented in any method or technology for long-term and short-term storage in any device, device or device. information such as computer-readable instructions, data structures, modules and sub-modules of programs or other data. Accordingly, the methods described herein can be encoded as executable instructions implemented in a computer-readable, hardware-based medium, including, but not limited to, a storage device and / or a memory device. When these instructions are executed by a processor, they cause the processor to implement at least a portion of the methods described herein. In addition, for the purpose of the present description, the term "computer-readable remanent media" covers all computer-readable hardware media, including, but not limited to, nonvolatile storage computing devices which, in no way limiting, volatile and nonvolatile media and removable and non-removable media such as a firmware, physical and virtual storage means, CD-ROMs, DVDs and any other digital source such as a network or the Internet, as well as digital means that will be developed in the future, the only exception being the propagation of a transient signal. For the purposes of this description, the term "real-time" includes the time when the associated facts and / or the timing of the predetermined measurement and data collection occur, the timing of the data processing and / or or the timing of the system's response to facts and the environment. In the embodiments described herein, these activities and facts occur in a substantially instantaneous manner. The wind farm control system described herein allows the use of acoustically measured acoustic pressure characteristics of a wind farm to facilitate the distinction between noise emanating from wind turbines and noise emanating from other sources. In particular, the systems and methods described herein use historical data to construct a probabilistic model for distinguishing noise from wind turbines from non-wind turbine noise. The history data includes noise recording segments of various durations that include recorded acoustic pressures as a function of time. Some of the noise segments will include acoustic pressures that will be substantially substantially noise produced by wind turbines, with only a small amount of contamination from non-wind turbine ambient noise, ie, these noise segments. will be "substantially uncontaminated" (in the present description, the terms "uncontaminated" and "undisturbed" will be used interchangeably). Alternatively, some of the noise segments will contain acoustic pressures with a high level of ambient noise contamination that does not emanate from wind turbines and will therefore be substantially "contaminated" (in the present description, the terms "contaminated" and "disturbed" will be used interchangeably). In this way, the systems and methods described herein are designed to improve the distinction between wind turbine noise and non-wind turbine noise, using a built-in probabilistic model linking these variations in acoustic characteristics to a wind turbine. the sound pressure. These variations in sound pressure measurements may be variations in the overall noise level and changes in the partial contributions between noise emitted by wind turbines and non-wind turbine noise, variations in the overall noise level. can be relatively constant. Therefore, the systems and methods described herein are designed to contribute to typification of contaminated and uncontaminated sound segments in acoustic pressure measurements. More specifically, the systems and methods described herein are designed to model acoustic characteristics associated with the wind farm and use the model to classify into uncontaminated and contaminated noise segments sound / noise recordings created as a result of acoustic pressure measurements. Noise segments indicating time slots where noise recordings are contaminated by extraneous noises from non-wind turbines, that is, time slots with a false sound pressure level (NPA) are rejected, that is to say marked as not acceptable. Foreign noise sources not emitting wind turbines include, but are not limited to, contributions from automobiles, aircraft and birds. As a result, the overall recorded acoustic history of the wind farm will be such that non-noise contributions are not associated with filtered wind noise on subsequent noise recordings. In this way, the methods and filtering algorithms described herein automatically differentiate between noise emanating from a wind turbine and foreign contributions.
[0006] In certain other possible embodiments, the systems and methods described herein are optionally adapted to correlate acoustic pressure variations, ie acoustic pressure (AdB), to known sources and facts by means of Probabilistic model built to make statistical estimates of sound pressure variations. On the other hand, in some embodiments, instead of just signaling the contaminated noise segments, the measured acoustic pressures can be corrected based on the estimates of AdB resulting from the modeled correlations. These embodiments include the elements necessary to create continuous statistical estimates of acoustic pressure, which facilitates obtaining more continuous feedback information in situations where long-term ambient noise contaminations are present, for example rain, vegetation noise, trains, road traffic and aircraft or short-term contaminations, eg insects and birds. FIG. 1 is a block diagram of an example of a computing device 105 that can be used to facilitate the operation of a plurality of wind turbines (not shown in FIG. 1) by means of a wind farm control system (FIG. not shown in Figure 1) installed at least partially in the computing device 105. More particularly, the computing device 105 creates acoustic feature models associated with a wind farm and uses the model to classify into uncontaminated and contaminated recorded sound segments. sound / noise recordings generated from acoustic pressure measurements. The computing device 105 includes a memory device 110 and a processor 115 that cooperates with the memory device 110 to execute instructions. In some embodiments, the executable instructions are stored in the memory device 110. The computing device 105 is configurable to perform one or more of the operations described herein by programming the processor 115. For example, the processor 115 may be programmed by encoding a processor. operation in the form of one or more executable instructions and storing the executable instructions in the memory device 110. In the exemplary embodiment, the memory device 110 consists of one or more devices that allow the storage and extracting information such as executable instructions and / or other data. The memory device 110 may include one or more computer-readable media.
[0007] The memory device 110 may be designed to store operating measurements including, in no way limiting, a history of acoustic pressures identified with noise emanating from wind turbines and noise not emanating from wind turbines, recordings of sound / noise and associated acoustic feature values, and any other type of data. Furthermore, the memory device 110 contains, in no way limiting, sufficient data, algorithms and instructions to facilitate the creation of acoustic feature models associated with a wind farm and uses the model to classify the records. sound / noise resulting from acoustic pressure measurements of uncontaminated and contaminated noise for specific sources of acoustic contamination or to access acoustic pressures induced by ambient noise sources.
[0008] In some embodiments, the computing device 105 also includes sufficient computer-readable / executable instructions, data structures, program modules, and program submodules to receive other data associated with value measurements. surveyed at another wind farm and other wind turbine systems to facilitate the overall operation of the wind farm. In some embodiments, the computing device 105 includes a presentation interface 120 coupled to the processor 115. The presentation interface 120 provides a user 125 with information such as a user input interface and / or an alert. In some embodiments, the presentation interface 120 includes one or more display devices. In some embodiments, the presentation interface 120 presents an alert associated with the wind farm control system being evaluated, including using a human-machine interface (HMI) (not shown in Figure 1). In some embodiments, the computing device 105 also includes a user input interface 130. In the exemplary embodiment, the user input interface 130 is coupled to the processor 115 and receives inputs made by the user. A communication interface 135 is coupled to the processor 115 and is adapted to communicate with one or more other devices such as a sensor or other computing device 105, and to perform input and output operations with respect to these devices while operating as an input channel. The communication interface 135 may receive data from one or more remote devices and / or transmit data to one or more remote devices. For example, a communication interface 135 of a first computing device 105 may transmit an alert to the communication interface 135 of another computing device 105. In some embodiments, the communication interface 135 is a radio interface . Figure 2 is a block diagram of a portion of a wind farm control system 200 operable to monitor and control at least a portion of a wind farm 300. In some embodiments, the control system 200 of wind farm also has sufficient computer-readable / executable instructions, data structures, program modules, and program submodules to receive other data associated with measured values at another wind farm and others wind turbine systems to facilitate the overall operation of the wind farm 300. Alternatively, the wind farm control system 200 is an autonomous system. Alternatively, another possibility is that the wind farm control system 200 is any computer system capable of monitoring portions of the wind farm 300 and creating noise models for the park 300. In the exemplary embodiment, the wind farm control system 200 comprises at least one central processing unit (CPU) 215 designed to execute control algorithms and control logic. The CPU 215 may be coupled to other devices 220 through a network 225. In some embodiments, the network 225 is a radio network. Referring to FIGS. 1 and 2, the CPU 215 is a computing device 105. In the exemplary embodiment, the computing device 105 is coupled to the network 225 through the communication interface 135. In another possible embodiment, the CPU 215 is integrated in other devices 220. The CPU 215 interacts with a first operator 230, for example, in a non-limiting manner, using a user input interface 130 and / or 120. In one embodiment, the CPU 215 provides the operator 230 with information about the wind farm 300, including acoustic pressure measurements or NPAs. Other devices 220 interact with a second operator 235, for example, in a non-limiting manner, using the user input interface 130 or the presentation interface 120. For example, other Devices 220 provide the operator 235 with alerts and / or other operating information. For the purposes of this description, the term "operator" covers any person associated with any title to the operation and maintenance of wind farm 300, including, in no way limiting, personnel working in teams, maintenance technicians and the supervisors of the facilities. In the exemplary embodiment, the wind farm 300 has one or more monitoring sensors 240 coupled to the central unit 215 by at least one input channel 245. The monitoring sensors 240 collect measurements of which, in no way limiting, acoustic pressures emanating from parts of the wind farm 300. The monitoring sensors 240 may also collect other operating measurements including, in no way limiting, speeds and directions of the wind in parts of the wind farm 300. In a repeated manner, for example periodic, continuous and / or on demand, the monitoring sensors 240 transmit readings of operating measurements at the time of the measurements. The CPU 215 receives and processes the readings of operating measurements. This data is transmitted via the network 225 and is accessible to any device capable of accessing the network 225 including, in no way limiting, desktops, laptops and personal digital assistants (PDAs) (none of them). not shown). Figure 3 is a schematic view of an example of a turbine generator 301 usable in the wind farm 300, both of which can be monitored and controlled using the wind farm control system 200 (shown in FIG. Figure 2). In the exemplary embodiment, the turbine generator 301 is a horizontal axis wind turbine. Alternatively, the turbine 301 may be a vertical axis wind turbine. The wind turbine 301 comprises a mast 302 extending from a support surface 304, a nacelle 306 mounted on the mast 302 and a rotor 308 mounted on the nacelle 306. The rotor 308 has a rotating hub 310 and a plurality of blades 312 In the exemplary embodiment, the rotor 308 has three rotor blades 312. Alternatively, the rotor 308 has any number of rotor blades 312 enabling the turbine generator 301 to operate as described herein. In the exemplary embodiment, the mast 302 is made of steel tubes and comprises a cavity (not shown in FIG. 3) extending between the support surface 304 and the nacelle 306. Alternatively, the mast 302 is any mast for the turbine generator 301 to function as described herein, including, but not limited to, a lattice mast. The height of the mast 302 has any value enabling the turbine generator 301 to operate as described herein.
[0009] The blades 312 are arranged around the rotor hub 310 to help turn the rotor 308, thereby transforming the kinetic energy of the wind 324 into usable mechanical energy, and then into electrical energy. The rotor 308 and the nacelle 306 are rotated on a yaw axis 316 around the mast 302 to control the perspective of the blades 312 with respect to the wind direction 324. The blades 312 are fitted on the hub 310 by mounting a root 320. The load transfer zones 322 comprise a hub load transfer zone and a blade load transfer zone (neither of which shown in Figure 3). The induced loads in the blades 312 are transmitted to the hub 310 via the charge transfer areas 322.
[0010] Each of the blades 312 also includes a blade tip portion 325. In the exemplary embodiment, the blades 312 have a length of 50 meters (m) (164 feet (ft)) and 100 m (328 ft), however these parameters do not constitute any limit for the present invention. Alternatively, the blades 312 may be any length allowing the wind turbine to operate as described herein. When the wind 324 strikes each of the blades 312, lift forces (not shown) of blades are induced on each of the blades 312 and the rotor 308 is rotated about the axis of rotation 314 as the pressure increases. acceleration of blade tip portions 325. A pitch angle (not shown) of the blades 312, i.e., an angle which determines the perspective of each of the blades 312 with respect to the wind direction 324, can be modified by a timing adjustment mechanism ( not shown in Figure 3).
[0011] In particular, increasing a stall angle of a blade 312 reduces a percentage of surface 326 exposed to the wind 324, and conversely, reducing a stall angle of a blade 312 increases a surface percentage 326. 324. For example, a blade pitch angle of about 0 degrees (sometimes referred to as a "power position") exposes a large percentage of a blade surface 324 to the wind 324, which causes forces to lift a first value to practice on the blade 312. Similarly, a blade pitch angle of about 90 degrees (sometimes called a "flag position") exposes the wind 324 a percentage well smaller of the surface 326 of the blade, which causes lift forces of a second value to be exerted on the blade 312. The first value of the lift forces caused to be exerted on the blades 312 is greater than the second value of the lift forces brought to practice on the blades 312, so that the va Their lift forces are directly proportional to the surface 326 of the blades exposed to the wind 324. Consequently, the value of the lift forces caused to be exerted on the blades 312 is related to the value of the angle of pitch of the blades. . In addition, as the speed of the tip portion 325 of a blade increases, an amplitude (not shown) of acoustic emissions (not shown in Figure 3) of the blade 312 increases. Conversely, as the speed of the tip portion 325 of a blade decreases, an acoustic emission amplitude of the blades 312 decreases. Therefore, the amplitude of the acoustic emissions of the blades 312 has a known relationship with a rotational speed of the blade tip portions 325, which ordinarily increases with a power of about 5/2 of the arrival speed and the amplitude of the acoustic emissions of the blades 312 has a known relationship with the pitch angle of the blades. The pitch angles of the blades 312 are adjusted about a pin axis 318 for each of the blades 312. In the exemplary embodiment, the pitch angles of the blades 312 are individually controlled. Alternatively, the pitch of the blades 312 can be controlled collectively. According to yet another possibility, the pitch of the blades and the speed of the blades 312 can be modulated to reduce the acoustic emissions. Preferably, a wind turbine 301 can be controlled to reduce potential acoustic emissions using a local controller (not shown) or remotely using a remote controller (not shown) to reduce noise. Noise reduction is usually accompanied by a decrease in annual energy production (EAP) because, for example, a reduction in the fixed-torque rotational speed directly reduces the output power. Figure 4 is a schematic view of a wind farm 300 which includes a plurality of wind turbines 301. In the exemplary embodiment, the wind turbines 301 are substantially similar to each other. Alternatively, the different wind turbines are made of various models. On the other hand, according to another possibility, rather than a plurality of wind turbines 301, the system 200 can be used with a wind farm 300 composed of a single wind turbine 301. The wind farm control system 200 comprises a plurality of devices. external noise meters, or microphones 350 which are an example of surveillance sensors 240. The microphones 350 are placed at predetermined locations in the park 300. The system 200 includes any number of microphones 350 deployed according to any which orientation allows the system 200 to operate as described herein. The microphones 350 are coupled to the CPU 215 (shown in Figure 2) via the input channel 245 (shown in Figure 2) which includes, in no way limiting, input-output lines Microphones 350 produce electronic signals (not shown) that are substantially representative of acoustic emissions or sound impressions (partially represented by isolines 351) emanating in the form of combined acoustic contributions from wind turbines as each of the blades 112 rotates. around the axis 114 (all shown in Figure 3). The electronic signals produced by the microphones 350 also contain contributions of ambient noise, that is to say non-wind turbine related foreign noises emanating from sources of which, in no way limiting, contributions from automobiles, aircraft and birds. In the exemplary embodiment, the microphones 350 produce signals (not shown) and transmit to the CPU 215 signals substantially representative of characteristics of broad bands and / or narrow bands of acoustic emissions emanating from each of the blades 112 of which , but in no way limiting, approximate values of frequency and amplitude. Acoustic measurements are usually acoustic pressures measured in a classical manner in pascals (Pa), which are converted to logarithmic scale and measured in decibels (dB). In the exemplary embodiment, acoustic measurements, i.e. acoustic pressure measurements of the wind farm 300 are recorded, which creates a temporal noise recording. The relationship generated between the acoustic pressure measurements with respect to the time of the plurality of acoustic pressure measurements of the noise recordings at least partially creates a history of the acoustic pressures. In some embodiments, these noise recordings are characterized as small fragments of acoustic signals transmitted through the memory device 110 and of short duration, that is, instead of being stored for a long time, as a result of the extraction of the acoustic characteristics data, these noise records are either eliminated or overwritten shortly after the data extraction. Acoustic pressure measurements may include amplitude modulated waves having a carrier frequency component and a modulation envelope component. The additional data collected includes, but is not limited to, power generation, loads, wind speed and wind direction, all in terms of time. In the exemplary embodiment, the wind farm 300 has a perimeter 352 and the microphones 350 are disposed within the perimeter 352, perimeter 352 and out of the perimeter 352. Alternatively, any combination of these provisions with respect to perimeter 352 and wind turbines 301 is used. In some embodiments, some of the wind turbine blades 301 of some wind turbines 301 may align with the microphones 350 in a manner that contributes to benefit from a more pronounced Doppler shift of the noise produced by the blades 312. when the Doppler effect is strong, that is to say when a microphone 350 is close to the plane of the rotor. Other blades 312 on other wind turbines 301 may not have such alignment with any of the microphones 350. Therefore, the determination of the wind turbine 301 that contributes to the overall noise level is facilitated. Since this analysis depends on the direction of the yaw movement, these microphones 350 will be limited in collecting information about the Doppler shifts of the blade tones. These unaffected Doppler shift sounds can be attributed to the noise of the machine, and these sounds can be matched to the expected harmonized frequencies associated with the rotor 308 speed (shown in Figure 3). The ambient noise is therefore probably other recorded sounds. Since the effect of Doppler shift depends on the direction of yaw movement, i.e. depends on rotor 308 with respect to yaw axis 316 (shown in FIG. 3), it will not always be applicable. . However, the arrangement of a microphone 350 may be adapted to a preferred direction. In some embodiments, the microphone 350 is radio frequency and is designed for one-way communication or two-way communication. The wind farm 300 also has a plurality of critical receiving points 354. These critical receiving points 354 are associated with locations where the overall noise levels associated with the wind farm 300 must meet regulatory requirements. Figure 5 is a schematic view of an example of a wind farm noise monitoring and control system 400 usable with the wind farm control system 200. In the exemplary embodiment, the system 400 includes microphones 350 disposed near the wind turbines 301 of the wind farm 300 described above. System 400 is part of the 200 wind farm control system. Alternatively, the wind farm noise monitoring and control system 400 is part of any other system (s) regardless of the architecture of the wind farm control system 200. .
[0012] On the other hand, in the exemplary embodiment, the wind farm control system 200 includes a wind farm controller 402 and a plurality of wind turbine controllers 404 coupled to the fleet controller 402. The fleet controller 402 mainly controls each of the wind turbines 301 by means of the wind turbine automatons 404, for example, in a non-limiting manner, by regulating the speed of rotation (number of revolutions per minute, or rpm) and the pitch angles of the blades 312 around the stall axis 318 (all shown in FIG. 3) as a function of the regulation of the general noise levels for the wind farm 300 based on the analyzes made by the CPU 215. In some embodiments, the wind farm control system 200 does not include the wind farm controller 402 and communicates directly with the wind turbine controllers 404. Furthermore, in the exemplary embodiment, each microphone 350 is part of a microphone station 406 which also includes a microphone controller 408 coupled to the microphone 350. The microphone controller 408 carries the acoustic pressure measurements made by the microphone. the microphone 350 to the CPU 215 for the analyzes described below. In the exemplary embodiment, the microphone station 406 comprises at least a portion of the computing device 105 (shown in FIG. 1), for example the memory device 110 and the processor 115 (both shown in FIG. ). These microphone stations 406 are designed to perform at least a portion of the methods described herein, including, but not limited to, time signal processing to obtain the relationship between acoustic pressure measurements and time. In some embodiments, the microphone station 406 is also configured to perform at least one noise filtering by a change estimate in the recorded noise data, for example, and in a non-limiting manner, by estimating noise values. AdB noise and a noise segment rating in a "good" or "bad" category. Partially processed data is transmitted to the CPU 215. Such methods are described in more detail below. In some other embodiments, a microphone 350 records only the measurements of the NPA and transmits the raw data to the CPU 215, for example, and in a non-limiting manner, a conventional microphone 350 with a preamplifier (not shown) in the microphone station 406 coupled by cable to data acquisition equipment (not shown) associated with the CPU 215. In addition, in the exemplary embodiment, the microphone stations 406 are coupled to the CPU 215 by a network 410 monitoring, control and data acquisition (SCADA). The SCADA 410 network facilitates the transmission to the CPU 215 of other data concerning the wind turbines and the wind farm. According to another possibility, or in addition to the SCADA network 410, the microphone stations 406 are coupled to the CPU 215 via a data transmission cable 412. Furthermore, according to one possibility or in addition to the SCADA network 410 and of the cable 412, the microphone stations 406 are coupled to the CPU 215 via a radio data transmission network 414 which comprises a plurality of antennas 416 facilitating radio frequency (RF) communications 418 and / or communication over the Internet 420, in particular through a cloud 422. In some embodiments, these methods may include, in no way limiting, specific RF communication protocols with special transceivers, on current bands, communications local radio network (WLAN) or communications over a network of general-purpose packet radio (GP) cellular telephones RS) / 3G.
[0013] In the exemplary embodiment, CPU 215 receives from all microphone stations 406 acoustic pressure measurements and any data thereof that are partially processed. The CPU 215 also determines the contribution to the total acoustic pressure measurements, ie the noise emanating from the wind turbines 301, by using a probabilistic model containing acoustic characteristics to filter noise that is not caused by a wind turbine. 301. The use of acoustic characteristics facilitates a better distinction between sources of noise and also facilitates the construction of a correlation (by gathering historical data) between changes in characteristics and measured acoustic pressures so that noises are corrected with greater precision. The CPU 215 furthermore executes at least a part of the methods described herein including, in no way limiting, the processing of time signals in order to obtain the relationship between the acoustic pressure measurements and the time. On the other hand, in some embodiments, the CPU 215 performs at least some noise filtering by estimating variations in the recorded noise data, for example, and in a non-limiting manner, by estimating the noise AdB values. and sorting the noise segment into the "good" or "bad" category. These methods are described in more detail below. In addition, the CPU 215 calculates operating values for the low noise operation (FBR) mode for at least some of the wind turbine 300 wind turbines 300, which mode reduces noise while limiting the decreases in annual energy production. (EAP). Figure 6 is a view illustrating an example of noise level variations likely to be associated with the wind farm 300 (shown in Figure 4). Figure 6 includes a simple wind turbine noise collection diagram 450 301, collected by microphones 350. Figure 6 also includes an associated curve 452 of sound pressure level (NPA) with respect to time, where a y axis 454 unitless represents sound pressure measurements of noise (on a logarithmic scale of the NPA, in dB) and an axis x 456 without unit represents time. The wind turbines 301 emit wind turbine noise 458 described above, which is transmitted by propagation in the atmosphere 460 to modify the noise 462 of wind turbines that has propagated, namely turbine noise 458 with propagation variations. . During propagation 460, the ambient noise 464, both continuous (for example, rain, nature and wind) that is intermittent (for example, aircraft and automobiles), and sometimes called disturbing noise, is added to produce a total measurement 466 of acoustic pressures. In general, disturbances by ambient noise can be severe. As a result, the wind farm noise control and limitation system 400 determines acoustic characteristics to classify parts of the total NPA into disturbed noise and undisturbed noise in the history, to determine changes in the total NPA due to a change of sources, a change in propagation and changes in the contribution of ambient noise, and in an ANPA value and possibly facilitate the determination of various types of disturbance sources by correlation with variations in characteristic values, which helps to improve the value of ANPA. On the other hand, it is possible to detect whether excessive amplitude modulation occurs and the corresponding reductions in the permissible NPA can be used as other input data for the noise limitation of a wind turbine. Models can be derived by site-specific learning, in some embodiments it is possible to transfer the model from one site to another and, in some embodiments, a database of characteristics and correlations is used. produced by synthesis.
[0014] Fig. 7 is a view illustrating an example of a probabilistic acoustic model, for example, and in no way limiting, a decision tree model 500 for determining acceptable records and unacceptable records using the monitoring system. and noise limitation 400 (shown in Figure 5). The decision tree model 500 determines the acceptability and unacceptability of noise records by a series of filtering steps. In this way, the decision tree model 500 is represented in the form of an inverted triangle where the unacceptable records (too much noise not coming from a wind turbine) and the acceptable records are distinguished from each other essentially. by a "yes / no" threshold analysis.
[0015] Fig. 8 is a flowchart illustrating an exemplary method 600 for operating the wind farm 300 (shown in Fig. 3) using the wind farm noise monitoring and control system 400 (shown in Fig. 5). . Referring to Figs. 7 and 8, in the exemplary embodiment, a plurality of acoustic pressure measurements of the wind farm 300 are recorded 602, which creates, 604, a plurality of noise records. In particular, the acoustic measurements, ie the acoustic pressure measurements of the wind farm 300 are recorded, which creates a noise recording with temporal characteristics, that is to say creates a relationship of the measurements of acoustic pressures over time for the plurality of acoustic pressure measurements. The time characteristics are facilitated either by microphone stations 406 (shown in FIG. 5) or by CPU 215. The relationship created with regard to acoustic pressure measurements with respect to time for the plurality of acoustic pressure measurements on noise recordings at least partially creates a history of temporal acoustic pressures. Acoustic pressure measurements include amplitude modulated waves having a carrier frequency component and a modulation envelope component. In this way, a total measurement 466 of acoustic pressures is created by or transmitted to the CPU 215 (shown in FIG. 5) via any architecture allowing the system 400 to operate as described herein. The total acoustic pressure measurement 466 is recorded to create at least one sound / noise archive 501 analyzed through the decision tree model 500. In addition, in the exemplary embodiment, values for a plurality Acoustic characteristics associated with the sound / noise archives 501 are calculated, 606. In this way, a plurality of statistical and non-statistical algorithms covering at least a portion of a predetermined frequency interval are executed, 608. non-limiting examples of non-statistical acoustic quantities the level weighting, where frequencies or moments of certain parts of the determined spectra are more heavily weighted than others, and where the determined spectra are determined in octave bands or bands 1st octave bands, such as 1/3 octave bands or narrow band spectra, with their power ances and their respective spectral densities. Statistical algorithms include fluctuation analysis, asymmetry analysis, standard deviation of the flow frequency band, approximation analysis and entropy analysis. In addition, any statistical analyzes that make it possible to operate the system 400 and the decision tree model 500 described here are used which, in a non-limiting manner, the flattening, the zero crossing rate, the cepstrum (transformed inverse Fourier of the logarithmic spectrum), Mel frequency scale (MFCC) cepstral coefficients and brightness. The processing of sound / noise recordings 501 is carried out by blocks of time segments of variable duration, that is to say, and in no way limiting, from about 10 milliseconds (ms) to several hours. , in order to capture the different characters of the ambient noise. For example, the passage of a train takes much longer than a gunshot and the overflights by the aircraft include frequencies much lower than those of insect stridulations. Ideally, the choice of acoustic characteristics for analysis facilitates the specific determination of different sources of disturbance. Further, in the exemplary embodiment, the probabilistic acoustic model, for example, and in a non-limiting manner is created, the decision tree model 500 of the wind farm 300, including the characteristic values. modeled acoustics present in it. To facilitate the distinction between wind turbine noise and non-turbine noise in total acoustic pressure measurements 466, the acoustic characteristics of sound / noise recordings 501 are used to construct a probabilistic model, for example, and in a non-limiting manner, the decision tree model 500 linking these changes in characteristics to changes in the total NPA to allow classification into disturbed acoustic segments (noise not emanating from wind turbines) and undisturbed acoustic segments (noise emanating wind). In this way, the history of the temporal acoustic pressure containing the acoustic recordings of the acoustic pressure measurements as a function of time comprises amplitude-modulated waves having a carrier frequency component and a modulation envelope component. execution, 608, of the plurality of statistical and non-statistical algorithms described above and in more detail hereinafter comprises an analysis of the carrier frequency component and / or the modulation envelope component, the frequency component. carrier being filtered or unfiltered. In addition, in the exemplary embodiment, the probabilistic acoustical model 500 of the wind farm 300 comprises one or more detection algorithms programmed in the CPU 215, that is to say implemented in the CPU. 215. The CPU 215 is designed to execute the detection algorithms in the model 500 so that a first contribution to the noise of the wind farm associated with at least one running wind turbine 301 and a second contribution of wind power noise associated with the sources other than wind turbines are distinguished in this one. The detection algorithms that at least partially define the model 500 are trained based on the noise measurement data from the wind farm 300 or with measured noise measurement data from a generic site. In addition, the learning periods may comprise a single learning cycle of the record analysis or may include a plurality of cycles for grasping, in a non-limiting manner, the seasonal effects of the weather and the changes in Physical and operational aspect of the wind turbines 301, the park 300 and the system 400. Whatever the source of the noise measurement data, the learning cycle of the model 500 includes the estimation of values of the difference in the measurement measurements. total acoustic pressures (A sound pressures) for sources other than wind turbines. In some embodiments, additional data, for example, but in no way limiting, rotor speed, wind speed, and stall angle, measured or calculated / estimated directly, are also provided to the 500 model. to capture specific conditions that may be associated with changes in acoustic pressure measurements. In this way, particular choices of acoustic characteristics for acoustic analysis facilitate the improvement of the determination of changes in the spectral shapes due to multiple contributions, which further enhances the independence of operation of noise-reduced wind turbines vis-à-vis -vis other sources of noise. The additional data further facilitates the distinction between a first contribution to wind turbine noise associated with running wind turbines 301 and a second contribution to wind farm noise associated with sources other than wind turbines. The greater the difference between the acoustic characteristics of the Model 500 between the noise emanating from wind turbines and the noise emanating from wind turbines, the greater the certainty of distinguishing noise emanating from wind turbines from noise that does not emanate wind turbines during subsequent noise recordings. In addition, in the exemplary embodiment of the model 500 and its learning, the acoustic pressures measured in the noise record segments are taken in dB to create the history data library to build the probabilistic model 500 in order to to distinguish the noise emanating from wind turbines from noise emanating from wind turbines. As described above, the history data includes noise recording segments of various durations, which contain the recorded acoustic pressures as a function of time. Some of the recorded noise segments will contain acoustic pressure measurements with a large contribution from the noise produced by wind turbines and with only a small contribution from contamination by ambient noise not emanating from wind turbines. Alternatively, some of the acoustic segments will contain acoustic pressures that include a large contribution from ambient noise contamination that does not emanate from wind turbines, of which segments or sources of noise that create contamination are noisier than wind turbines. In the exemplary embodiment, the model 500 comprises the undisturbed acoustic record segments, i.e., especially the wind turbine noise and acceptable, and noise record segments that include a contamination. by noise from ambient sources sufficient to be classified as disturbed and not acceptable. These distinctive noise segments are then used to define the probabilistic model 500 using a plurality of statistical analyzes. In this way, although A's of acoustic pressures are calculated during learning of the model 500, estimates of A of acoustic pressures are determined during the learning succeeding phase, i.e. subsequent measurements of the acoustic pressures. In some embodiments, to further facilitate the distinction between the first contribution to acoustic pressure measurements with respect to the second contribution to sound pressure measurements from sources other than wind turbines, noise from sources other than wind turbines is recorded over a plurality of time periods described above, for subsequent analysis. Further, in the exemplary embodiment, the probabilistic model 500, which relies on the distinguishing characteristics between the undisturbed noise and the disturbed noise depending on the selected acoustic characteristics, is used to analyze the next noise record made during the operation of the wind farm 300. During the "passage" of the subsequent noise records in the 500 model, the noise segments indicating time slots during which the noise recordings are contaminated by foreign noise that does not emanate from the noise record. wind turbines, ie time segments with skewed sound pressure, are rejected, that is, marked as not acceptable. Sources of foreign noise from non-wind turbines include, but are not limited to, contributions from automobiles, aircraft and birds. Therefore, with respect to all the acoustic data of wind farm 300 subsequently recorded, the contributions of the foreign noise will not be associated with the noise of the wind turbines separated by filtering indicating time slices of relevant noise. In this way, the model 500 automatically executes the filtering methods and algorithms described in more detail below to establish a difference between the noise emitted by wind turbines and the estimated foreign contributions to indicate the noise segments subsequently recorded, these segments being consisting of noise emanating substantially from wind turbines with a noise of contamination lower than a predetermined threshold. In some embodiments of the model 500, the contributions of the foreign noise sources can be determined with confidence, at least with estimated values of AdB so that the noise sources are identified and the associated correlated noise segments can easily be provided with an indication of known contaminations, which makes the filtering process faster. Therefore, in these embodiments, the probabilistic acoustic learning model 500 of the wind farm 300 includes a distinction of the second contribution to acoustic pressure measurements originating from sources other than wind turbines. In this way, the contributions of the second contribution to acoustic pressure measurements are calculated and compared with differences in the measurements of total acoustic pressures (A of acoustic pressures) and statistical estimates of A of acoustic pressures for the model 500 are produced. . In addition, in some embodiments, rather than a mere indication of disturbed noise segments, the estimated values of AdB in subsequent acoustic pressure measurements are corrected, it being understood that the removal of known foreign sources of noise in noise ensures more accurate dB values of wind turbines 301. In this way, irrespective of the fact that the changes in acoustic pressure measurements result from the known operation of wind turbines 301 or known ambient sources, the distinction of the second contribution to Acoustic pressures from sources other than wind turbines include linking the changes in the values calculated for the plurality of acoustic characteristics of the model 500 to acoustic signatures associated with known noise sources present in the probabilistic acoustic model 500 of the wind farm 300. In certain forms As a result, the Model 500 is continuously updated with new data based on the analysis of subsequent noise records. These new data can be added to the temporal acoustic pressure history by filtering the second contribution to acoustic pressure measurements by an indication of the parts of the history of temporal acoustic pressures that contain the noise levels due to ambient sources exceeding predetermined thresholds. In the exemplary embodiment, the statistical algorithms include fluctuation analysis, asymmetry analysis, standard deviation analysis of the flow frequency band, approximation analysis, and entropy analysis. Each is discussed in more detail below. According to another possibility, any statistical analysis, in any order, allowing the operation of the system 400 and the decision tree model 500 described here are used, of which, in no way limiting, the flattening, the zero crossing rate, the cepstrum (inverse logarithmic spectrum Fourier transform), the Mel-scale frequency cepstral coefficients (MFCC) and the brightness. The model 500 comprises a first filtering module, namely a module 502 of fluctuation means that analyzes a first acoustic characteristic. The module 502 compares averages of the amplitude fluctuations of the sound / noise record 501 around at least a predetermined average of amplitude with the values of the psychoacoustic model of the fluctuation force present in the model 500. As an exemplary embodiment, the average fluctuations of the amplitudes of noise over the entire frequency band are analyzed. Overall, the characteristics can also be evaluated on one or more predetermined frequency bands smaller than the entire frequency interval. The frequency bands are each defined by a first predetermined lower frequency value and a second predetermined larger frequency value. Any frequency band is used to operate the system 400 and model 500 described herein. For example, and in no way limiting, in some embodiments, the first and second frequencies are respectively zero and positive at infinity. Overall, the frequency bands that facilitate the distinction between noise emanating from a wind turbine and noise that does not emanate from a wind turbine are chosen based on, and in no manner limited to, the type, the model and the manufacturer. wind turbines analyzed. In some embodiments, the determination of the appropriate frequency bands may require a first choice and then some refinement to focus on a frequency band, or a plurality of frequency bands, for the analyzes. For example, and in no way limiting, a lower frequency band, a high frequency band and a medium frequency band are selected. In some embodiments, some of these frequency bands have a predetermined overlap. In addition, for example, and in a non-limiting manner, it is possible to use nested frequency bands, i.e. a wider frequency band contains one or more narrower frequency bands nested in this one. One or more predetermined fluctuation thresholds are also chosen, these for example being a lower limit of 113 for the average fluctuation. In module 502, a relationship is determined, 612, between the values calculated for the first acoustic characteristic, that is, the fluctuations with respect to the noise recording and the modeled fluctuations in the probabilistic acoustic model 500 of the wind farm 300. Model 500 contains the values of the most representative acoustic characteristics of undisturbed noise, that is, a first contribution to the acoustic pressure history measurements associated with wind turbines 301 of wind farm 300.
[0016] If the relation, ie the comparison between the fluctuation values, indicates that the second contribution to sound / noise recordings 501 from sources other than wind turbines has been extremely contaminated, i.e. When disturbed sound / noise recordings 501, a first contribution to sound / noise recordings 501 by wind turbines 301 is distinguished, 614, from the second contribution to sound / noise recordings 501 from sources other than wind turbines. If it is found that the segment of its correspondent is disturbed, the segment of the sound recording is marked by an indicator and will not continue until the next module (this point is discussed below). However, if the determined relationship indicates that the second contribution to sound / noise recordings 501 from sources other than turbines has not been extremely contaminated, i.e. did not disturb the sound / noise recordings 501, the corresponding segment of the acoustic recording is transmitted to the next module (this point is discussed below). In this way, the contaminated segments of the sound / noise recording 501 are filtered to be separated from the non-contaminated parts of the sound / noise recordings 501. The model 500 also comprises a second filtering module, namely a module of asymmetry 504 which analyzes a second acoustic characteristic. The module 504 compares values of the third central moment, i.e., the average asymmetry of the amplitudes of the sound / noise record 501, with the values of the average asymmetry amplitude present in the model 500. In the exemplary embodiment, the asymmetry of the noise amplitudes around an average amplitude value is representative of symmetry, or asymmetry of the noise amplitude values around the average value. predetermined. A substantially symmetrical distribution will have an asymmetry value close to zero. In module 504, a relationship is determined, 612, between the calculated values for the second acoustic characteristic, that is, the asymmetry with respect to noise recording and the modeled asymmetry in the probabilistic acoustic model. 500 of the 300 wind farm. Model 500 contains the values of the most representative acoustic characteristics of undisturbed noise, that is, a first contribution to historical sound pressure measurements associated with wind turbines 301 of wind farm 300 In the exemplary embodiment, a predetermined average asymmetry threshold value of about 15,000 is used. In this way, any average asymmetry value greater than about 15,000 indicates disturbed noise segments. Alternatively, any average asymmetry threshold value enabling the operation of the system 400 and the decision tree model 500 described herein is used. If the relation, ie the comparison between the asymmetry values, indicates that the second contribution to the recording of sound / noise 501 by sources other than wind turbines is extremely contaminated, that is to say ie, disturbed sound / noise recording 501, a first contribution to sound / noise recording 501 by wind turbines 301 is distinguished, 614, from the second contribution to sound / noise recording 501 by sources other than wind turbines. If it is found that the corresponding noise segment is disturbed, the segment of the noise recording is marked by an indicator and does not continue until the next module (this point is discussed below). However, if the determined relationship indicates that the second contribution to sound / noise recording 501 from sources other than wind turbines has not been extremely contaminated, ie did not interfere with the recording of sound / noise 501, the corresponding segment of the noise recording is transmitted to the next module (this point is discussed below). In this way, the contaminated segments of the sound / noise recording 501 are filtered, 616 to be separated from the non-contaminated parts of the sound / noise record 501. The model 500 further comprises a third filter module, which is namely the modulus 506 of spectral flow frequency standard deviation which analyzes a third acoustic characteristic. The module 506 compares values of the standard deviation of the distances between the spectrum of successive slices of sound / noise records 501 with the standard values of the standard deviation of the distances between the spectrum of successive slices present in the model 500. In the exemplary embodiment, the distances between the spectrum of successive slices is representative of a rate of variation of the spectrum of sound / noise recording 501 by comparing a slice with a previous slice. Large spectral variations reveal a disturbance. In the exemplary embodiment, the frequency band for the analysis is from about 800 Hz to about 1600 Hz. Alternatively, any frequency bands that allow the operation of the system 400 and of the model 500 described here. In the module 506 is determined, 612, a relationship between the values calculated for the third acoustic characteristic, i.e., the standard deviation of flux with respect to the noise record and the standard deviation of the modeled flow. in the probabilistic acoustic model 500 of wind farm 300. Model 500 contains the values of the most representative acoustic characteristics of undisturbed noise, ie a first contribution to historical sound pressure measurements associated with wind turbines 301 of the wind farm 300. In the exemplary embodiment, a predetermined standard deviation of flux values of less than about 0.25 is used. In this way, any standard deviation of flux values below about 0.25 reveals disturbed noise segments. Alternatively, any flow standard deviation values that allow the operation of the system 400 and the decision tree model 500 described herein are used.
[0017] If the relationship, ie the comparison of the standard deviation of flux values, indicates that the second contribution to sound / noise recording 501 from sources other than wind turbines has been extremely contaminated, that is, has contaminated the sound / noise record 501, a first contribution to sound / noise recording 501 by wind turbines 301 is distinguished, 614, from the second contribution to acoustic pressure measurements from other sources than wind turbines. If it is found that the associated noise segment is disturbed, the segment of the noise record is marked with an indicator and does not continue until the next module (this point is discussed below). However, if the determined relationship indicates that the second contribution to sound / noise recording 501 from sources other than wind turbines has not been extremely contaminated, ie did not disrupt the recording of sound / noise 501, the associated segment of the noise record is transmitted to the next module (this point is discussed later). In this way, the contaminated segments of the sound / noise recording 501 are filtered, 616, to be separated from the uncontaminated parts of the sound / noise record 501. The model 500 also comprises a fourth filter module, which is namely an approximation module 508, which analyzes a fourth acoustic characteristic. The module 508 compares values of the approximation, that is to say a sensory dissonance related to beat phenomena when pairs of sinusoidal acoustic signals are close to each other in frequencies, of the recording. 501 with the model values of the approximation values present in the model 500. In the exemplary embodiment, the amplitude peaks in the predetermined frequency band are produced by different noise sources and, if these amplitude peaks are sufficiently large and exhibit constructive interference or resonance, they may have a significant impact on sound / noise recording 501. In particular, the dissonant beats may disturb the noise recordings. In the exemplary embodiment, the frequency band for the analysis is from about 1600 Hz to about 3200 Hz. Alternatively, any frequency bands that allow the system 400 to operate are used. of the model 500 described here. In the module 508 is determined a relation 612 between the values calculated for the fourth acoustic characteristic, that is to say the approximation from the sound / noise record 501 and the modeled approximation in the probabilistic acoustic model. 500 of the 300 wind farm. The 500 model contains the values of the most representative acoustic characteristics of undisturbed noise, ie a first contribution to historic acoustic pressure measurements associated with wind turbines 301 of wind farm 300 In the exemplary embodiment, a predetermined maximum value of approximation of less than about 3.25 is used. In this way, any approximation values below about 3.25 indicate disturbed noise segments. Alternatively, any approximation values that allow the operation of the system 400 and the decision tree model 500 described herein can be used. If the relation, ie the comparison between the approximation values, indicates that the second contribution to the recording of sound / noise 501 by sources other than wind turbines has become extremely contaminated, that is, that is, disturbed the recording of sound / noise 501, a first contribution to the recording of sound / noise 501 by wind turbines 301 is distinguished, 614, from the second contribution to acoustic pressure measurements by sources other than wind turbines . If it is found that the corresponding noise segment is disturbed, the segment of the noise recording is signaled by one indicator and does not continue until the next module (this point is discussed below). However, if the determined relationship indicates that the second contribution to sound / noise recording 501 from sources other than wind turbines did not become extremely contaminated, that is, did not disturb the recording of sound / noise 501, the corresponding segment of the noise recording is transmitted to the next module (this point is discussed below). In this way, the contaminated segments of the sound / noise recording 501 are filtered, 616, to be separated from the non-contaminated parts of the sound / noise record 501. The model 500 also includes a fifth filter module, which is namely an entropy module 510 which analyzes a fifth acoustic characteristic. The module 510 compares values of the average entropy, i.e., the randomness related to the recurrent amplitude frequency, of substantially similar height values that may reveal measurable noise emanating from a turbine or a turbine. a source other than a turbine. This measurable noise can have a significant impact on sound / noise recording 501. In the exemplary embodiment, the frequency band for analysis is about 200 Hz to about 400 Hz. any frequency bands for operation of system 400 and model 500 described herein are used. In the module 510 is determined, 612, a relationship between the calculated values for the fifth acoustic characteristic, i.e. entropy; based on noise recording and modeled entropy in Probabilistic Acoustic Model 500 of Wind Farm 300. Model 500 contains the values of the most representative acoustic characteristics of undisturbed noise, a first contribution to historical acoustic pressures associated with wind turbines 301 of the wind farm 300. In the exemplary embodiment, a predetermined entropy value of less than about 0.56 is used. In this way, any entropy values below about 0.56 indicate disturbed noise segments. Alternatively, any approximation values for the operation of the system 400 and decision tree model 500 described herein are used. If the relation, ie the comparison between the entropy values, indicates that the second contribution to the recording of sound / noise 501 by sources other than wind turbines has contaminated too much, that is to say that is, disturbed the recording of sound / noise 501, a first contribution to the recording of sound / noise 501 by wind turbines 301 is distinguished, 614, from the second contribution to acoustic pressure measurements by sources other than wind turbines . If it is found that the corresponding noise segment is disturbed, the segment of the noise recording is signaled by an indicator and will not be saved. However, if the determined relationship indicates that the second contribution to sound / noise recording 501 from sources other than wind turbines has not been extremely contaminated, ie did not interfere with the recording of sound / noise 501, the corresponding segment of the noise recording is saved. In this way, the contaminated segments of the sound / noise recording 501 are filtered, 616, to be separated from the non-contaminated parts of the sound / noise record 501. When the disturbed recordings are signaled by an indicator, they are transmitted to a queue 512 of unacceptable records. Undisturbed records transmitted via the model 500 and not flagged are sent to a queue 514 of acceptable records. Unreasonable records in queue 512 may be completely rejected or may be stored in a separate library for future analysis. Acceptable records in queue 514 may be added to the historical sound pressure measurement library, or sound / noise recordings to enhance acceptable noise levels associated with the first contribution to pressure measurements. In certain circumstances, strong noise contamination is present and a sufficiently accurate estimate of the noise emanating from wind turbines is not possible. The CPU 215 is designed to return to a default operating method, or the sound at the receiving location can be extrapolated from the history of previous measurements. Previous measurements can be assisted by known changes in atmospheric conditions such as, in no way limiting, wind speed, direction, shear and temperature. In addition, recorded noise emission measurements can be used as servo signals for wind farm control optimization algorithms. In addition, an intermediate solution is that the detection of the blade passing frequency (FPP) for estimating the speed and improving the noise recognition of a wind turbine is possible, in particular to distinguish Similar low-frequency disturbances such as a far-distant aircraft based on fast completion Fourier Transforms (FFTs) with zeros or autocorrelation of the signal itself or its envelope. In this way, amplitude modulation abnormalities, affecting the FPP, could also be detectable. The distinction of the second contribution to acoustic pressure measurements from sources other than wind turbines includes the analysis of the total acoustic pressure measurements for a predetermined period of time after the production of the noise recording. With regard to the time scales of the estimation and the reactions to the changes, the algorithms present in the 500 model do not need to be very fast, because the regulations can simply require a respect over a longer period, for example, multiple hours. A duration of time slots, for example, in a non-limiting manner, from 10 seconds to 60 seconds facilitates obtaining long-term averages of filtered acoustic pressures. In addition, a distinction can be made almost in real time to facilitate the real-time control of the wind turbines 301 in order to limit the overall noise levels of the wind farm 300. In particular, the first and second parts of the noise recordings can be distinguished. from each other so that only the first contribution to the sound recordings corresponding to the noise emanating from wind turbines 301 is retained and used for the real-time control of the wind turbines 301. During the learning period of the model 500, the processing speed may be lower than the processing speed once the 500 model has completed learning. Once the learning is complete, the speed of the model 500 is higher because there is no need to give a history for the learning and the means of treatment can be mainly devoted to the analysis succeeding the noise recordings. In the exemplary embodiment, and in no way limiting, steps 610 and 612 of the method are used to teach the model 500 as part of the usual learning phase for probabilistic models. FIG. 9 is an example of a graphical view (graph) 700 of evaluation of data as a function of the mean asymmetry (y-axis 702) and the mean fluctuation (x-axis 704) using the architecture 400 (FIG. shown in Figure 5) of a wind farm noise monitoring and control system with, installed in the system 400, a probabilistic acoustic model other than the decision tree model 500 (shown in Figure 7). Similarly, Figure 10 is an example of a graphical view (graph 710) of estimates of noise data as a function of standard deviations of flux (y-axis 712) and of the mean fluctuation (x-axis 714) using of architecture 400 (shown in FIG. 5) of a wind farm noise monitoring and limitation system with, installed in the system 400, a probabilistic acoustic model other than the decision tree model 500 (shown in FIG. 7). Other possible embodiments of probabilistic acoustic models include, but are not limited to, a neural network, a carrier vector regression (SVR) model, or a combination of both. For purposes of this description, the terms "artificial neural network (ANN)" and "neural network (RN)" are intended to mean any computer-implemented programs and computer systems modeling the complex relationships between input and output. outputs or find combinations in the data. Moreover, in the sense of the description, the RNAs and RNs are adaptive systems that modify their structure according to external or internal information that circulates in the network during a learning phase. In addition, within the meaning of the present description, the terms "carrier vector regression model (SVR)" and "carrier vector machine (MVS)" are intended to refer to any computer-implemented classification methods. and computer-based ones that construct hyperplanes in a multidimensional space for analyzing data, recognizing combinations, classifying and sorting similar attribute data into a same set of defined groups, typifying and sorting data with similar attributes and / or different sets of defined groups and develop the ability to predict this classification and / or typing after "learning" using learning data. In the exemplary embodiment, the y-axis 702 (shown in FIG. 9) for the average asymmetry ranges from about 0.5 * 104 to about 4.5 * 104 in increments of about 0. , 5 * 104. The y-axis 712 (shown in Figure 10) for the standard flow deviation ("stdev flux") ranges from about 0.03 to about 0.27 in increments of about 0. .05. The x-axis 704 (shown in Fig. 9) and x-axis 714 (shown in Fig. 10) are substantially similar and range from about 120 to about 245 in increments of about 20. In some forms of In view of the precise values sought, a calibration factor is provided for converting acoustic pressure units to pascal. On the other hand, in the exemplary embodiment, the graphs 700 and 710 define a matrix of rectangles that are assigned a color code or a pattern code (as shown in FIGS. 9 and 10). The colors and patterns indicate a set of rectangles of undisturbed and disturbed areas where the overall average difference in sound pressures is determined from noise records using established acoustic models of wind farm 300 (shown in Figure 3). 4) trained for RN and RVS models. In particular, FIG. 9 includes a legend 705 of patterns and FIG. 10 includes a legend 715 of patterns that have both patterns representing a variable scale between a lower perturbation and a larger perturbation. When the A's of acoustic pressure values are calculated (estimated), it appears that the graphs 700 and 710 have a smaller A of acoustic pressure values ("good") (indicating undisturbed noise recordings) and A of higher ("bad") sound pressure values (indicating disturbed noise records ordinarily not associated with wind turbine noise). The "good" areas are areas 706 and 716 respectively for Figs. 9 and 10, and the "bad" areas are 708 and 718 respectively. The "good" areas 706 and 716 represent combinations of acoustic characteristics that correspond to disturbances. relatively low noise from a wind turbine, that is, acceptable areas of sound / noise recordings. Likewise, the "bad" zones 708 and 718 represent the combinations of acoustic characteristics that correspond to relatively high disturbances by noise not emanating from wind turbines, that is to say unacceptable zones of sound recordings. /noise.
[0018] The use of these methods facilitates the discovery and reporting of noise segments extremely contaminated by ambient noise. Moreover, the use of these methods facilitates the correction of the acoustic pressure values measured for the A of estimated acoustic pressure values. In addition, the use of these methods facilitates the execution of an all-or-nothing filtering algorithm for acoustic signals. In addition, the use of these methods facilitates the use of complex interrelations of acoustic characteristics to analyze noise recordings. In addition, the use of these methods facilitates continuous statistical estimates of A of acoustic pressure values. Fig. 11 is an example of a graphical view (graph 750) of comparison of various acceptable record determination methods and unacceptable recordings using the surveillance system architecture 400 (shown in Fig. 5). and wind farm noise limitation. The graph 750 includes an γ axis 752 representing a percentage of accepted noise segments on the total number of recorded noise segments, ranging from 0% to 100% in increments of 10%. The curve 750 also includes an x axis 754 representing the set of A's of estimated or measured acoustic pressure values, ranging from 2 dB to 14 dB, in increments of 2 dB. The graph 750 further comprises a decision tree curve 756 representing the percentage of accepted recordings as a function of the A of determined acoustic pressure values. The graph 750 also includes a neural network curve 758 representing the percentage of accepted recordings as a function of the A of determined acoustic pressure values. The neural network model is more efficient in marking by an indicator of the noise segments recorded for the A's of acoustic pressure values determined as the values increase, signaling disturbed noise segments. Figure 12 is an exemplary configuration 800 of a database 802 present in a computing device 804, as well as other associated computer components, which can be used to operate the wind farm 300 (shown in Figure 4) described right here. The database 802 is coupled to several separate components in the computing device 804, which perform specific tasks. In the exemplary embodiment, the computing device 804 may be the computing device 105 (shown in Figure 1) or the CPU 215 (shown in Figure 2). The computing device 804 is configured to interface with a human operator 805 of the system.
[0019] In the exemplary embodiment, the database 802 contains data 806 on wind turbines, data 808 on the wind turbine control system, noise data 810 from wind turbines, and data 812 on the wind turbine. noise not coming from wind turbines. The wind turbine data 806 includes information such as configuration data, for example, and in no way limiting, the number of blades of the wind turbines and the speed of rotation. The data 808 on the wind turbine control system includes information associated with the architecture of the wind turbine control system 200, including, in a non-limiting manner, the wind farm noise monitoring and limiting system 400 . The noise data 810 from wind turbines include the data associated with the first contribution to the wind farm noise by the operating turbine (s) 301 and the noise data 812 not from wind turbines include data. associated with the second contribution to wind farm noise from sources other than turbines, as described herein. The computing device 804 includes the database 802, as well as data storage devices 814. The computing device 804 also includes a wind turbine automaton component 816 for performing steps 602 to 616 of the method (shown in Figure 8). The computing device 804 also includes a medium fluctuation module component 818 (first filter module), an asymmetry module component 820 (second filter module), a flow standard deviation module component 822 (third module filtering device), an approximation module component 824 (fourth filter module) and an entropy module component 826 (fifth filter module) all designed to perform steps 606 to 616 of the method. The computing device 804 further includes a processing component 828 which contributes to the execution of computer executable instructions associated with the wind farm noise control and limitation system 400, the method 600 and the configuration 800 described herein. In addition, any statistical analysis of any acoustic characteristics that allows the operation of the system 400, the method 600 and the configuration 800 described here in particular, in a non-limiting manner, flattening, Zero crossing rate, cepstrum (inverse Fourier transform of logarithmic spectrum), Cepstral coefficients of Mel scale (MFCC) and brightness.
[0020] The above described wind farm control system allows the use of acoustically measured sound pressure characteristics of a wind farm to facilitate the distinction between noise emanating from wind turbines and noise emanating from other sources. In particular, the systems and methods described herein use historical data to construct a probabilistic model to distinguish between noise from wind turbines and noise from non-wind turbines. The history data includes noise record segments of various durations, which include recorded acoustic pressures as a function of time. Some of the noise segments contain acoustic pressures that are mostly noise produced by wind turbines, with only a small amount of contamination from ambient noise that does not emanate from wind turbines, ie, these noise segments are substantially uncontaminated or undisturbed. Alternatively, some of the noise segments will contain acoustic pressures that will have a high level of contamination by ambient noise that does not emanate from wind turbines and is therefore highly contaminated or disturbed. In this way, the systems and methods described herein are designed to contribute to a better distinction between noise emitted by wind turbines and non-wind turbine noise by using a probabilistic model constructed for these changes in noise characteristics. compared to the sound pressure. These changes in acoustic pressure measurements may be changes in the overall noise level and offsets in the partial contributions between noise from wind turbines and noise from non-wind turbines while changes in the overall noise level can be relatively constant. Therefore, the systems and methods described herein are designed to facilitate the classification of contaminated and uncontaminated noise segments of the measured acoustic pressures. More specifically, the systems and methods described herein are designed to model acoustic characteristics associated with the wind farm and to use the model to classify sound / noise recordings produced from measured acoustic pressures into uncontaminated and contaminated noise segments. . In addition, the wind farm control system described here contributes to improving the enslavement of the wind turbines of the wind farm. In particular, the embodiments described herein help to improve the control of the reduced noise operating mode (FBR) using real-time, day-to-day, and single-season change determinations. the other, in acoustic characteristics in addition to the usual environmental characteristics, for example, wind speed, wind direction and air density. As such, the embodiments described herein help to improve power generation efficiency and increase annual power generation (EAP) by taking into account time-related variables, such as changes in wind turbine configuration. , including interruptions in wind turbine operation and fouling and blade erosion. In addition, the greater distinction between disturbed and undisturbed acoustic pressures contributes to the use of narrower margins than regulatory parameters, which further enhances electricity production. In addition, power generation by wind farms is increasing due to the increase in the density of the wind turbine population on sites with limited space. An example of a technical effect of the methods, systems and devices described herein includes: (a) the correction of acoustic pressures measured in a wind farm based on model correlations based on probabilistic assessments of long-term contamination and short duration by noise that does not emanate from wind turbines; and / or (b) improving the reduced noise operating mode (FBR) control through real-time, day-to-day, and season-wise change determinations. other in acoustic characteristics in addition to the usual environmental characteristics; and / or (c) improving the efficiency of electricity production and increasing annual energy production (EAP) by taking into account time-related variables such as wind turbine configuration changes of which shutdowns of wind turbines and blade contamination and erosion; and / or (d) reinforcing the distinction between contaminated and non-contaminated noise segments, which facilitates the use of narrower margins relative to regulatory parameters, further enhancing electricity production; and / or (e) improving the distinction between noise emanating from wind turbines and noise from non-wind turbines using a probabilistic model constructed for these changes in noise characteristics with respect to wind turbines. sound pressure; and / or (f) the classification of measured acoustic pressures into contaminated and uncontaminated noise segments.
[0021] Examples of embodiments of methods, systems and devices for operating wind farms are not limited to the specific embodiments described herein but, instead, system components and / or process steps can be used. independently and separately from other components and / or steps described herein. For example, the methods can also be used in combination with other systems requiring real-time control based on ambient acoustic conditions in real time as well as historical acoustic characteristics correlated with park history conditions and processes, and are not limited to implementation only with the systems and methods described herein. In contrast, the exemplary embodiment can be implemented and used in conjunction with a large number of other applications, equipment and systems that can benefit from this acoustic monitoring and regulation. Although specific aspects of various embodiments of the invention may be shown in some drawings and not others, this is only for convenience. According to the principles of the invention, any detail of a drawing may be cited and / or claimed in combination with any detail of any other drawing. Some embodiments involve the use of one or more electronic or computer devices. These devices usually comprise a processor or a PLC such as a general purpose CPU (CPU), a graphics processor (GPU), a microcontroller, a computer processor with reduced instruction set (RISC), an integrated circuit for application specific (ASIC), a programmable logic circuit (CLP) and / or any other circuit or processor capable of performing the functions described herein. The methods described herein may be encoded as executable instructions implemented in a computer-readable medium including, in a non-limiting manner, a storage device and / or a memory device. These instructions, when executed by a processor, cause the processor to implement at least a portion of the methods described herein. The examples above are just examples and are not intended to limit the definition or meaning of the term.

权利要求:
Claims (20)
[0001]
REVENDICATIONS1. A computer-implemented method (600) for operating a wind farm (300) using a computing device (105; 215; 804) including at least one processor (115; 816; 826) coupled to a device memory device (110), the wind farm having at least one turbine (301) running, said method comprising: recording (602) a plurality of wind farm sound pressure measurements, including producing a recording noise ; calculating (606) values for a plurality of acoustic characteristics associated with the noise recording; determining (612) a relationship between the calculated values for the plurality of acoustic characteristics and the modeled acoustic characteristic values present in a probabilistic acoustic model (500) of the wind farm; and the distinction (614), according to the determined relation, of a first contribution to noise recording by the running wind turbine (s) with a second contribution to noise recording by sources other than wind turbines.
[0002]
The method (600) according to claim 1, further comprising generating (610) the probabilistic acoustic model (500) of the wind farm (300) having a first contribution associated with the wind turbine (s) (301) in operation. and a second contribution associated with sources other than wind turbines.
[0003]
The method (600) of claim 1, wherein computing (606) values for a plurality of acoustic features comprises performing (608) a plurality of statistical and non-statistical algorithms on at least a portion of the a predetermined frequency interval.
[0004]
The method (600) of claim 3, further comprising generating (604) a time-based acoustic pressure measurement relationship for the plurality of acoustic pressure measurements.
[0005]
The method (600) of claim 4, wherein the relationship of acoustic pressure measurements with respect to time comprises amplitude modulated waves having a carrier frequency component and a modulation envelope component, the execution (608) of a plurality of statistical and non-statistical algorithms comprising analyzing the carrier frequency component and / or the modulation envelope component.
[0006]
The method (600) of claim 1, wherein discriminating (614) a first input to the noise recording comprises filtering (616) the second contribution to noise recording by sources other than wind turbines based on the noise record of the wind farm (300).
[0007]
The method (600) of claim 6, wherein filtering (616) the second contribution to noise recording includes signaling, by indicators, portions of the noise record.
[0008]
The method (600) of claim 1, wherein discriminating (614) a first contribution to noise recording includes distinguishing the second contribution to noise recording from sources other than wind turbines for a plurality of times.
[0009]
The method (600) of claim 8, wherein distinguishing the second contribution to noise recording from sources other than wind turbines comprises analyzing the second contribution to the noise recording during a lapse of time. predetermined time and after the production of the noise recording.
[0010]
The method (600) of claim 8, wherein distinguishing the second contribution to noise recording from sources other than wind turbines comprises: calculating inputs of the second contribution to noise recording in differences in measurements of total acoustic pressures (A of acoustic pressures); and the production of statistical estimates of A of acoustic pressures.
[0011]
The method (600) of claim 8, wherein distinguishing the second contribution to noise recording from sources other than wind turbines comprises using known sources of the second contribution to the noise recording to to learn the probabilistic acoustic model (500) of the wind farm (300).
[0012]
The method (600) of claim 8, wherein distinguishing the second contribution to noise recording from sources other than wind turbines includes establishing a link between changes in the values calculated for the plurality. acoustic characteristics and acoustic signatures associated with known noise sources present in the probabilistic acoustic model (500) of the wind farm (300).
[0013]
13. A wind farm (300) comprising: a plurality of wind turbines (301) including a running wind turbine; a microphone (350) disposed near said wind farm; and a computing device (105; 215; 804) coupled to said microphone, said computing device comprising a processor (115; 816; 826) and a memory device (110) coupled to said processor, said computing device being adapted to: produce (604 ) recording noise by recording a plurality of acoustic pressure measurements of said wind farm using said microphone; calculating (606) values for a plurality of acoustic characteristics associated with the noise recording; determining (612), using the computing device, a relationship between the calculated values for the plurality of acoustic characteristics and the modeled acoustic characteristic values present in a probabilistic acoustic model (500) of said wind farm; and distinguishing (614), from the predetermined relationship, a first contribution to noise recording by said running wind turbine with a second contribution to noise recording by sources other than wind turbines.
[0014]
The wind farm (300) of claim 13, wherein said microphone (350) comprises a microphone station (406) comprising at least a portion of said computing device (105; 215; 804).
[0015]
15. Wind farm (300) according to claim 13, wherein said microphone (350) comprises a plurality of microphones disposed in a perimeter (352) and / or on the perimeter and / or outside the perimeter of said wind farm.
[0016]
The wind farm (300) according to claim 13, wherein said microphone (350) is aligned with said running wind turbine (301), facilitating the Doppler shift analysis of at least a portion of the wind turbine (301). noise recording.
[0017]
The wind farm (300) of claim 13, wherein said microphone (350) is radio frequency and is designed for one-way communications and two-way communications.
[0018]
The wind farm (300) according to claim 13, wherein said computing device (105; 215; 804) comprises a noise monitoring and noise control station (400) adapted to regulate the operation of said wind turbine (301) in operation. at least partially according to the predetermined relationship.
[0019]
19. A wind farm (300) according to claim 13, said computing device (105; 215; 804) being furthermore designed to produce the probabilistic acoustic model (500) of said wind farm comprising a first contribution associated with said wind turbine (301) in March and a second contribution associated with sources other than wind turbines.
[0020]
20. Computer-operable single or multiple storage medium (110; 225; 814) in which computer-executable instructions are stored, the computer-executable instructions, when executed by at least one processor ( 110; 816; 826), causing the processor (s) to produce (604) a noise recording by recording a plurality of acoustic pressure measurements of a wind farm (300); calculating (606) values for a plurality of acoustic characteristics associated with the noise recording; determining (612) a relationship between the calculated values for the plurality of acoustic characteristics and the modeled acoustic characteristic values present in a probabilistic acoustic model (500) of the wind farm; and discriminating (614), from the determined relationship, a first contribution to noise recording from a running wind turbine (301) relative to a second contribution to noise recording from sources other than éoliennes.20

类似技术:

---

公开号 | 公开日 | 专利标题

---

FR3024502A1|2016-02-05|

---

Priyadarshani et al.2018|Automated birdsong recognition in complex acoustic environments: a review

---

US20060173623A1|2006-08-03|System and method for enhanced measure-correlate-predict for a wind farm location

---

US11100918B2|2021-08-24|Event sensing system

---

US20200233397A1|2020-07-23|System, method and computer-accessible medium for machine condition monitoring

---

EP3710693B1|2022-01-26|System and method for diagnosing a rotor unbalance of a wind turbine

---

EP3172548B1|2018-09-26|Method for detecting anomalies in a distribution network, in particular for drinking water

---

Krause et al.2020|Damage detection for wind turbine rotor blades using airborne sound

---

Rieutord et al.2021|Deriving boundary layer height from aerosol lidar using machine learning: KABL and ADABL algorithms

---

CN112727710B|2022-02-11|Wind field thunderbolt density statistical method and system based on audio signals

---

JP2019502859A|2019-01-31|Method for evaluating incident flow in rotor blade of wind power generation facility, method for controlling wind power generation facility, and wind power generation facility

---

EP3650687B1|2021-06-16|Method for determining an induction factor for a wind turbine provided with a laser remote detection sensor

---

Chen et al.2021|Parameterization of wind evolution using lidar

---

CN109933933B|2020-09-01|Noise treatment method and equipment

---

EP3489831A1|2019-05-29|Method and device for monitoring a data generator process of a metric for the prediction of abnormalities

---

EP3712621A1|2020-09-23|Method for predicting the wind speed in the rotor plane of a wind turbine provided with a sensor for remote laser detection

---

d N Santos et al.2022|Data-driven farm-wide fatigue estimation on jacket-foundation OWTs for multiple SHM setups

---

WO2021209726A1|2021-10-21|System for real-time recognition and identification of sound sources

---

JP4513796B2|2010-07-28|Abnormality monitoring device

---

WO2021205509A1|2021-10-14|Feature amount extraction device, time-sequential inference apparatus, time-sequential learning system, time-sequential feature amount extraction method, time-sequential inference method, and time-sequential learning method

---

FR3065804A1|2018-11-02|METHOD FOR EVALUATING THE ACOUSTIC IMPACT OF A VIRTUAL WIND FARM

---

CN113298134A|2021-08-24|BPNN-based remote non-contact health monitoring system and method for fan blade

---

Runkel et al.2021|The Handbook of Acoustic Bat Detection

---

Kokkinopoulos2020|aster R

---

Alías Pujol et al.2018|DYNAMAP: A low-cost WASN for real-time road traffic noise mapping

同族专利:

---

公开号 | 公开日

---

CA2898931A1|2016-01-31|

---

US20160032893A1|2016-02-04|

---

CA2898931C|2017-12-05|

---

US9347432B2|2016-05-24|

---

FR3024502B1|2021-12-17|

---

DE102015112594A1|2016-02-04|

引用文献:

---

公开号 | 申请日 | 公开日 | 申请人 | 专利标题

---

---

DE50008607D1|1999-06-10|2004-12-16|Aloys Wobben|WIND ENERGY SYSTEM WITH SOUND LEVEL CONTROL|

---

SE0200237D0|2002-01-29|2002-01-29|Abb Ab|Apparatus and method for operation of a power generating plant|

---

US20070031237A1|2005-07-29|2007-02-08|General Electric Company|Method and apparatus for producing wind energy with reduced wind turbine noise|

---

US20070050441A1|2005-08-26|2007-03-01|Step Communications Corporation,A Nevada Corporati|Method and apparatus for improving noise discrimination using attenuation factor|

---

US8277185B2|2007-12-28|2012-10-02|General Electric Company|Wind turbine, wind turbine controller and method for controlling a wind turbine|

---

EP2337952B1|2008-09-30|2012-11-07|Vestas Wind Systems A/S|Control of wind park noise emission|

---

US7988414B2|2008-10-20|2011-08-02|General Electric Company|Method and system for operating a wind turbine generator|

---

WO2010061255A2|2008-11-01|2010-06-03|Clipper Windpower, Inc.|Active blade pitch control for reduction of wind turbine noise or loads|

---

US9230219B2|2010-08-23|2016-01-05|Institute Of Nuclear Energy Research Atomic Energy Council, Executive Yuan|Wind energy forecasting method with extreme wind speed prediction function|

---

US20110223018A1|2010-12-21|2011-09-15|Prashant Srinivasan|Control System, Wind Farm, And Methods Of Optimizing The Operation Of A Wind Turbine|

---

CN103797244B|2011-08-16|2017-09-12|维斯塔斯风力系统集团公司|Noise acoustic monitoring system for wind turbine|

---

US8304926B2|2011-09-16|2012-11-06|General Electric Company|Wind turbine sound management|

---

DE102011086990A1|2011-11-23|2013-05-23|Wobben Properties Gmbh|Method for controlling a flight-obstruction lighting or a wind farm for carrying out such a method|

---

EP2599996B1|2011-12-02|2016-06-29|Vestas Wind Systems A/S|Controlling of noise emission of a wind park|

---

CN202746093U|2012-06-26|2013-02-20|华锐风电科技股份有限公司|Noise control device of wind driven generator|US9995277B2|2014-07-31|2018-06-12|General Electric Company|System and method for controlling the operation of wind turbines|

---

US10156224B2|2015-03-13|2018-12-18|General Electric Company|System and method for controlling a wind turbine|

---

DE102016011005A1|2016-09-14|2018-03-15|Senvion Gmbh|Method and system for analyzing a condition of a wind turbine|

---

US10733533B2|2017-03-07|2020-08-04|General Electric Company|Apparatus and method for screening data for kernel regression model building|

---

WO2018187859A1|2017-04-11|2018-10-18|Systèmes De Contrôle Actif Soft Db Inc.|A system and a method for noise discrimination|

---

CN110094300B|2019-01-24|2020-04-21|上海电力学院|Wind deviation correction method, device, equipment and medium for wind turbine generator|

---

DE102019104820A1|2019-02-26|2020-08-27|fos4X GmbH|Wind energy installation, sensor kit for a wind energy installation, system with a wind energy installation and an online-based storage and server service and method for sending data from a wind energy installation to an online-based storage and server service|

---

GB2584111A|2019-05-21|2020-11-25|Renewal Energy Systems Ltd|System and method for controlling the noise emission of one or more wind turbines|

---

CN110880046A|2019-11-06|2020-03-13|清华大学|Short-term unit information acquisition method and device, computer equipment and storage medium|

---

CN110838302B|2019-11-15|2022-02-11|北京天泽智云科技有限公司|Audio frequency segmentation method based on signal energy peak identification|

---

US11060504B1|2020-02-07|2021-07-13|General Electric Company|Systems and methods for continuous machine learning based control of wind turbines|

法律状态:
2016-07-26| PLFP| Fee payment|Year of fee payment: 2 |

---

2017-07-26| PLFP| Fee payment|Year of fee payment: 3 |

---

2018-06-21| PLFP| Fee payment|Year of fee payment: 4 |

---

2019-06-21| PLFP| Fee payment|Year of fee payment: 5 |

---

2020-06-23| PLFP| Fee payment|Year of fee payment: 6 |

---

2021-03-05| PLSC| Publication of the preliminary search report|Effective date: 20210305 |

---

2021-06-23| PLFP| Fee payment|Year of fee payment: 7 |

优先权:

---

申请号 | 申请日 | 专利标题

---

US14/447,848|US9347432B2|2014-07-31|2014-07-31|System and method for enhanced operation of wind parks|

---