• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    system and method for the automatic identification of a device or apparatus. a system for the automatic identification of a device or device (105a-105c) includes: at least one sensor (110) configured to be associated with the device or device to be identified, and to monitor a time evolution of at least a quantity electrical power consumption by the device or apparatus; an analyzer (110; 115; 125) in communication relation with at least one sensor and configured to receive reports from him (215) of the monitored electrical quantity, the analyzer being further configured to automatically identify the device or device being analyzed (220) at evolution in time of the monitored electrical quantity. the analysis includes; calculate a cross-correlation (530, 555) between the time evolution of the monitored electrical quantity and at least one reference standard representative of at least one sample device or apparatus; insert at least one sample device or apparatus into a list of candidate devices or apparatus in each case that a value related to the calculated cross-correlation is above a predetermined threshold; in the event that more than one sample device or device is included in the candidate list, identify the device or device by performing a selection between the candidate device or devices based on characteristic parameters related to the respective and different calculated cross correlations of said value.
    公开号:BR112013002490B1
    申请号:R112013002490
    申请日:2010-08-04
    公开日:2020-01-28
    发明作者:Borean Claudio；Elia Gabriele；Claps Leonardo
    申请人:Telecom Italia Spa；
    IPC主号:
    专利说明:

    “SYSTEM AND METHOD FOR THE AUTOMATIC IDENTIFICATION OF A DEVICE OR APPARATUS”
    Background of the Invention
    Field of Invention [001] The present invention generally relates to the field of home automation / home automation. More specifically, the present invention relates to a method and a related system allowing automatic recognition of electrical devices / appliances such as household appliances, and in particular, a method and system designed to enable automatic detection of devices / devices, for example , electrical devices / appliances connected to the power source, using a home area network, for example, and not limited to, a Wireless Sensor Network (WSN).
    Related Technique Overview [002] Home automation is an evolving field. In this context, several solutions have been and are being deployed for energy management purposes, and standards are emerging to cover application scenarios. Some of the known solutions exploit WSNs. As known, a WSN is a network of autonomous nodes spatially distributed with sensors to cooperatively monitor physical or environmental conditions, such as temperature, sound, vibration, pressure, movement, pollutants. WSNs are currently used in many industrial and civil application areas, including industrial process monitoring and control, machine health monitoring, environment and habitat monitoring, healthcare applications, home automation, traffic control. In addition to one or more sensors, each node in a WSN is equipped with a radio transceiver or other wireless communications device, a small microcontroller, and possibly a power source, usually a battery. A WSN is usually an ad-hoc network
    Petition 870190087388, of 9/5/2019, p. 9/60 / 24 wireless, meaning that each sensor supports multi-jump routing (multiple nodes can forward data packets to a base station).
    [003] GB 2451001 describes an intelligent measurement system using low power radio transmission; the system consists of several sensors that generate consumption signals and a central unit for display information and graphical data on measurement.
    [004] WO 2009/097400 exposes systems and methods to monitor and control the energy consumption of an energy consuming device. During the installation phase, the position information of smart sockets is fixed: this information is then used to process the energy signal, therefore.
    Summary of the Invention [005] The Applicant has observed that most known solutions require an initiation procedure that involves direct human intervention in order to specify which system the specific sensor node will monitor and control. This is believed to be a significant disadvantage, especially due to the fact that the average home user can be, and often in fact is, unaccustomed to technical tasks and thus can be discouraged.
    [006] The Claimant tackled the technical problem of inventing a solution that allows monitoring and possibly controlling devices / appliances, particularly although not limited to electrical devices / appliances, using a home area network, such as a WSN, without the need for direct or indirect human interaction, at least not in the system installation phase.
    [007] The Applicant has found a solution that enables an automatic procedure for the discovery of devices / devices, particularly devices / devices connected to the home area network, particularly WSN; taking advantage of this functionality, the solution according to the present invention avoids direct human intervention for
    Petition 870190087388, of 9/5/2019, p. 10/60 / 24 specify the devices / devices to be monitored and possibly controlled.
    [008] Essentially, in accordance with an aspect of the present invention, a system is provided for the automatic identification of a device or apparatus. The system includes:
    - at least one sensor configured to be associated with the device or device to be identified, and to monitor a time evolution of at least an electrical amount indicative of a power consumption by the device or device; and
    - an analyzer in communication relation with said at least one sensor and configured to receive reports of the monitored electrical quantity from it, the analyzer being furthermore configured to automatically identify the device or apparatus analyzing the evolution in time of the monitored electrical quantity.
    This analysis includes:
    - calculate a cross correlation between the evolution in time of the monitored electrical quantity and at least one reference standard representative of at least one sample device or apparatus;
    - insert at least one device or sample device in a list of candidate devices or devices in the event that a value related to the calculated cross-correlation is above a predetermined threshold;
    - if more than one sample device or device is included in the candidate list, identify the device or device by making a selection between the applicant devices or devices based on characteristic parameters related to the respective calculated cross correlations and different from said
    Petition 870190087388, of 9/5/2019, p. 11/60 / 24 value.
    [009] In particular, said at least one reference standard includes at least two reference standards, each corresponding to a respective observed evolution in time of the monitored electrical quantity in relation to a specific operational mode of the same device or sample device .
    [0010] Said at least one reference standard can be generated by averaging observed evolutions in time of the monitored electrical quantities of devices or devices, or by observing the evolution in time of the monitored electrical quantity of the device or device to be identified during a first cycle operation, or detecting characteristic events in the time observed in the monitored electrical quantity.
    [0011] In particular, detecting said characteristic events includes detecting peaks in the time evolution of the monitored electrical quantity.
    [0012] In one embodiment of the present invention, calculating said cross-correlation may include determining a time observation window and cross-correlating the monitored electrical quantity and at least one reference pattern in at least one time interval corresponding to the time window determined observation.
    [0013] Preferably, calculating said cross-correlation is repeated in at least two different time intervals of duration equal to the determined time window.
    [0014] Said time window of observation can, for example, be determined by detecting characteristic events in the evolution observed in time of the monitored electrical quantity.
    [0015] In one embodiment of the present invention, said analysis includes, before calculating the cross correlation, transforming the observed evolution in time of the monitored electrical quantity and at least one
    Petition 870190087388, of 9/5/2019, p. 12/60 / 24 reference standard in the frequency domain, and in which to calculate this cross-correlation is performed in the results of the transformation.
    [0016] In particular, said characteristic parameters can include at least one between an average value, maximum relative values, standard deviation.
    [0017] The system can advantageously be embedded in an intelligent receptacle by which the device or electrical device is connectable to an electrical receptacle.
    [0018] According to another aspect of the present invention, a method is provided for the automatic identification of a device or apparatus, the method including:
    - associate at least one sensor with the device or device to be identified, the sensor being configured to monitor a time evolution of at least an electrical quantity indicative of energy consumption by the device or device;
    - receive data on the monitored electrical quantity informed by at least one sensor, and automatically identify the device or apparatus analyzing the evolution in time of the monitored electrical quantity.
    [0019] This analysis includes:
    - calculate a cross correlation between the evolution in time of the monitored electrical quantity and at least one reference standard representative of at least one sample device or apparatus;
    - insert at least one device or sample device in a list of candidate devices or devices in the event that a value related to the calculated cross-correlation is above a predetermined threshold;
    in the event that more than one sample device or apparatus
    Petition 870190087388, of 9/5/2019, p. 13/60 / 24 is included in the candidate list, identifying the device or device by performing a selection among the candidate devices or devices based on characteristic parameters related to the respective calculated cross correlations and different from said value.
    [0020] Some of the advantages of the present invention are:
    - automated installation and network commissioning procedures without human interaction; the system is then “plug and play”, and the network initiation is done faster;
    - improved utility of the energy management solution enabled by the home area network;
    when smart sockets are used to measure the energy requested by devices, the method automatically recognizes an exchange between receptacles assigned to different devices.
    Brief Description of the Drawings [0021] These and other features and advantages of the present invention will be more readily understood after reading, together with the accompanying drawings, the following detailed description of some exemplary and non-limiting embodiments thereof. In the drawings:
    Figure 1 schematically shows a system according to an embodiment of the present invention;
    Figure 2 is a schematic flow chart of the main steps of a method according to an embodiment of the present invention;
    Figure 3 is a schematic flow chart of a first procedural part of a device / apparatus recognition process of the method of Figure 2, according to an embodiment of the present invention;
    Figure 4 is a schematic flowchart of a second part of the recognition process procedure, according to an embodiment of the present invention;
    Petition 870190087388, of 9/5/2019, p. 14/60 / 24 Figure 5 is a schematic flowchart of a third part of the recognition process procedure, according to an embodiment of the present invention;
    Figure 6 schematically describes a procedure for selecting a device from different possible candidates; and Figure 7 schematically shows an intelligent receptacle according to an embodiment of the present invention.
    Detailed Description of Embodiments of the Drawings [0022] Referring to the drawings, Figure 1 schematically shows an architecture of a system according to an embodiment of the present invention, particularly a system for administration and energy control of devices / devices, capable of automatically discover and detect devices and / or devices connected to an electricity distribution network (that is, in the described embodiment, devices / devices), based on a processing that, as will be described in detail later, explores measurements of the absorbed energy performed on the devices / devices themselves using sensors from a home area network, for example, a WSN.
    [0023] In Figure 1, reference numeral 100 denotes a house, for example, an apartment, in which electrical devices and / or appliances are deployed, by way of example, a personal computer 105a, a television set 105b and a washing machine clothing 105c; clearly, different and / or additional electrical devices / appliances can be installed in apartment 100, for example, a refrigerator, an oven, a dishwasher, a dryer, as well as devices / appliances that are not necessarily powered by electrical energy (eg thermal energy devices / appliances).
    [0024] The electrical devices / appliances 105a, 105b, 105c are connected to the AC receptacles of the house by smart receptacles 110,
    Petition 870190087388, of 9/5/2019, p. 15/60 / 24 each incorporating a sensor node from a home area network, preferably a wireless network, for example, a WSN (next, and unless otherwise stated, reference numeral 110 will be used equivalently to denote the smart receptacle or the sensor node built into it); the use of smart receptacles 110 including sensor nodes is advantageous because there is no need to have dedicated devices / devices already embedding sensor nodes, so that the present invention can also be practiced together with devices / devices already deployed; however, there is nothing to prevent the present invention from being exploited even if devices / devices with built-in sensor nodes are used. The sensors in the sensor nodes 110 are configured to perform measurements of power and electrical energy in the devices / devices that are connected to the respective smart receptacles; the sensors can include an integrated circuit operating the measurements, and a relay to manage the electrical energy to be distributed to the loads connected to the physical AC receptacle.
    [0025] In embodiments of the present invention, the home area network is, for example, as mentioned above, a WSN. Each sensor node 110 includes a wireless transceiver, adapted to communicate wirelessly with other WSN nodes based on a wireless communication protocol, for example, and not limited to, a ZigBee protocol, or other management protocols satisfactory mesh networks. Some or possibly each of the WSN nodes 110 are configured to act as relay nodes, that is, as wireless routers between the other WSN 110 nodes and a connection point 115 (furthermore, nothing prevents the point function connection is embedded in one of the WSN 110 nodes).
    [0026] Connection point 115 is configured to manage electrical devices / devices 105a, 105b, 105c and join, from WSN nodes
    Petition 870190087388, of 9/5/2019, p. 16/60 / 24 respective 110, information related to different devices / devices.
    [0027] Connection point 115 is also configured to communicate, over a communication network 120, as a telephone network (either mobile or not, or a combination of the two) and / or a packet data network such as Internet, with a remote platform 125, including one or more servers, possibly distributed; the remote platform 125 is configured to gather data sent by WSN nodes 110 to connection point 115 and store it in satisfactory databases 130 in order to make the stored data accessible by remote applications (access to data administered by the remote platform 125 by remote applications can, for example, run using high-level APIs (Application Program Interfaces).
    [0028] A method according to an embodiment of the present invention for automatic recognition (i.e., identification) of the type of electrical devices / apparatus connected to the electricity distribution network will now be described.
    [0029] Figure 2 is a schematic flowchart of the main phases of the method.
    [0030] The electrical devices / devices are connected by the user to the receptacles of the AC distribution network by the intelligent receptacles 110, which incorporate the sensor nodes (block 205). In the event that the device / device already incorporates a sensor node, the smart receptacle is not required and the device / device can be connected to the main AC network receptacle directly.
    [0031] In a configuration phase, the sensors on sensor nodes 110 are automatically configured by connection point 115 (block 210), for example using a wireless communication protocol, such as the ZigBee protocol; the configuration is aimed at enabling the sensor nodes
    Petition 870190087388, of 9/5/2019, p. 17/60 / 24
    110, and the sensors built into it, perform, on the associated electrical devices / appliances, measurements of instantaneous electrical energy consumption and absorbed energy. For example, sensor nodes 110 are configured in such a way as to make sensor nodes 110 inform the connection point 115 information related to the measured device / apparatus energy consumption and absorbed energy (block 215). The information procedure for sensor nodes 110 for connection point 115 can, for example, follow the following guidelines:
    • periodically report measurement data to connection point 115 after a maximum length of time (TMAX) elapsed between a previous measurement data report;
    • report extemporaneously to connection point 115 when the measured value of instantaneous energy and cumulative energy is above certain thresholds; such thresholds can be pre-stored on sensor nodes 110 or they can be configured via connection point 115 (either locally at the connection point or remotely by a host application using connection point 115 itself) in the configuration phase: in this In this case, the generic sensor node 110 reports to the connection point 115 a change in the value of the monitored quantity.
    [0032] According to an embodiment of the present invention, in order to avoid the need for a manual configuration of the association between the sensor nodes 110 and the respective devices / devices, a procedure is performed to automatically recognize the connected device / electrical device to the AC receptacle by the smart receptacle 110 containing the sensor nodes (block 220). In one embodiment of the present invention, the automatic device / apparatus recognition procedure can preferably include the following three phases (which are described in detail in the following):
    • training sets selection phase (block 220a);
    Petition 870190087388, of 9/5/2019, p. 18/60 / 24 • segmentation phase (block 220b);
    • wedding verification phase (block 220c).
    [0033] The device / device recognition procedure is repeated until the device / device is identified (blocks 225 and 230).
    [0034] Phase 220a for the selection of training sets can, according to an embodiment of the present invention, be performed as outlined in the flow chart of Figure 3. A training set can be defined in time as a reference standard for the evolution of energy consumption, corresponding to a known electrical device / appliance or class of electrical devices / appliances. For example, instantaneous energy over time can be chosen as the amount to be monitored and used to identify the devices / devices.
    [0035] According to the reporting procedure described above, the generic sensor node 110 is configured to report the measured data to the connection point 115 in the event of changes in the measured values above predetermined thresholds, or in any case, even when no change greater than the predetermined thresholds occurs, after the maximum TMAX time duration of a previous measurement data report is elapsed. In particular, the generic sensor node 110 measures the instantaneous power consumption of the associated device / apparatus. Thanks to this, the instantaneous energy measurements reported can advantageously be used by the connection point 115 (or the remote platform 125) to track the actual energy consumption curve of the device / electrical device concerned.
    [0036] It is observed that the procedure for reporting measurement data from sensor nodes to the connection point involves a reduction in the number of measurement samples available at the sensor nodes (the number of measurement samples taken by the sensor, for example a energy meter, is significantly higher than the number of samples
    Petition 870190087388, of 9/5/2019, p. 19/60 / 24 transmitted to connection point 115 by sensor node 110). This reduction in the number of measurement samples can be described as a noise disturbing the measurement of the monitored quantity, for example, instantaneous energy. Such noise can nevertheless be filtered out through the marriage verification process described later.
    [0037] A sliding time window of satisfactory time duration is used to process the incoming data about actual energy consumption measurements provided by the generic sensor node 110 for connection point 115. The data contained in the current time window (a “ of the data flow received at connection point 115) are compared to one or more training sets.
    [0038] According to an embodiment of the present invention, three options are provided for the selection of training sets:
    a) a (ie, unique) set of trainee (block 305): the data contained within the current time window are compared to a single training set; thus, it can be useful for those types of electrical devices / appliances that, in operation, substantially always exhibit the same behavior;
    b) family of training sets (block 310): in this case, the data contained in the current time window are compared to two or more, possibly several different training sets from a family of training sets, which can, for example, describe different behaviors of the electrical device / appliance (for example, a washing machine that can be programmed to operate according to different washing programs);
    c) “self-test” (block 315): in this case, the measured data collected in relation to the first operating cycles of the electrical device / apparatus (that is, the measured instantaneous energy data corresponding, and collected in relation to the first operating cycles) device / device
    Petition 870190087388, of 9/5/2019, p. 20/60 / 24 electrical) are taken as the training set, that is, as the reference standard to be used thereafter for comparisons with the data contained in the sliding time window.
    [0039] In cases a) and b), the nature of the source of the training set is then chosen (block 320). Possible sources of training set are:
    - real source: the training set to be used for device / device recognition is selected from training sets corresponding to possible different behaviors of known electrical devices / devices (block 325). For example, the training set can be chosen from a group of reference standards corresponding to different functions of known devices / devices, each describing a specific behavior of known devices / devices (for example, a low temperature washing cycle of a washing machine, an economical washing cycle for a dishwasher); these functions are obtained by observing and selecting the change over time of the instantaneous energy consumed by the device / device while operating according to a specific function;
    - simulated source: in this case, training sets are created by averaging the main characteristics of real behaviors (block 330). After several attempts at marriage using real training sets, the most successful ones are processed to create a new training set, being an average of them.
    [0040] In case c), the collected data are scanned, looking for the first peak in the instantaneous energy consumption curve (block 335). The data is scanned for example using the same technique as described in the following for the “event recognition” process in the
    Petition 870190087388, of 9/5/2019, p. 21/60 / 24 of segmentation 220b. Then, the energy consumption curve following the first energy peak is ‘‘ recorded ’until the measured instantaneous energy returns to zero for a certain time. The recorded data becomes the training set to be used later (block 340).
    [0041] Segmentation step 220b can, according to one embodiment of the present invention, be performed as outlined in the flowchart of Figure 4.
    [0042] The segmentation process performed in segmentation phase 220b is aimed at identifying the operating cycle of the device / electrical device connected to sensor node 110, enabling comparison between discrete sets of data that need to be classified correctly. In particular, the segmentation process aims to create the time window that will then be explored in the marriage phase.
    [0043] Firstly, the data received from sensor nodes 110 are gathered in a database (block 405); data can be collected locally at connection point 115 or remotely at remote platform 125; the processing of the joined data can be performed by the connection point 115 or by the remote platform 125. Before the data is processed, data relating to an observation time period are gathered; the observation time period can be configurable, and can be set, for example, to correspond to a day, a week or a longer period.
    [0044] Due to the different arrival times of each sample of measurement data, the collected data should be managed and put on the same time basis. This can be done by linearly interpolating (for example, performing a Spline interpolation using a polynomial of degree higher than 1) the data and arrival time, the former being described, for example, by a time stamp named on the received data ( block 410); in one embodiment of the invention, the interval between two samples of
    Petition 870190087388, of 9/5/2019, p. 22/60 / 24 successive data can be about 10 seconds, to avoid excessive growth of the data set.
    [0045] The time window to be used in the subsequent marriage phase is then chosen (block 415). The time window can be defined in two ways:
    a) using a sliding window, which creates a matching window for each interpolated sample;
    b) using an “event recognition” method, which allows recognizing the presence of an event and isolating its own time window in relation to this.
    [0046] The main difference between these two methods lies in the computational time, because the sliding window method asks to check the marriage continuously.
    [0047] Let case a) be considered. The length (width) of the training set is calculated (block 420). The width of the time window is defined by the length of the training set. Since the choice can be made among several possible training sets, each with its own length, the time windows created by the segmentation process may differ from case to case. The time window is then created: the time window is created for each sample, and has a width defined by the length of the training set (block 425).
    [0048] In case b), the main edges of the energy consumption curve for the device / device to be recognized are located (block 430). For this purpose, two thresholds can be defined:
    - an energy threshold: if the instantaneous energy absorbed by the device / device exceeds this value, the beginning of a leading edge is declared;
    - time threshold: if the energy consumption of the device / appliance remains above the energy threshold for a period of
    Petition 870190087388, of 9/5/2019, p. 23/60 / 24 time longer than the time threshold, a leading edge is recognized.
    [0049] A scan is then performed over the entire observation time period, looking for the leading edges using the previously defined thresholds. At the end of the process, a list of starting points for all the main borders is created. After each main edge detected, the end of the event is determined (block 435). For this purpose, a search is performed to find the first sample that satisfies the following two thresholds:
    - energy threshold: if the device's energy falls below this value, the end of the event is declared;
    - time threshold: if the device's power remains below the energy threshold for longer than the time threshold, the end of the event is recognized.
    [0050] The time window is then created (block 440), based on the positions of the detected main edges and the respective ends of the events.
    [0051] According to an embodiment of the present invention, the wedding verification phase 220c is based on Bayesian decision theory. The formula for decision is:
    P (r I tf 1)> / ε
    P (r | ífO) / <
    where P (r | H1) is an a posteriori probability of hypothesis H1, P (r | H0) is an a posteriori probability of hypothesis H0. The left relation of inequality is called “plausibility function”, and it can be considered as the distance between the two hypotheses, H1 and H0. If the plausibility function is greater than the decision threshold ε, hypothesis H1 is chosen, otherwise hypothesis H0 is chosen.
    [0052] The method described below aims to identify the device / device that minimizes the distance between its own
    Petition 870190087388, of 9/5/2019, p. 24/60 / 24 instantaneous energy consumption x (t) and the y (t) curve defined by the selected training set. The formula for the distance used is the Euclidean distance:
    D = f. Mt) - y (t)] Λ = Ex + E, - 2 x (t) y (t) dt [0053] The integral in the previous formula is developed in its components using the integral transformation linearity property (f (x + y) = f (x) + f (y) and f (ax) = af (x)); Ex and Ey are the energy of each function (instantaneous energy consumption curve of the device / device to be recognized and the training set, respectively), while the last sum in the formula on the right can be seen as the cross correlation in a window where the two functions overlap. Cross-correlation is a function that is useful to provide a measure of the affinity of two different signals and the delay needed for them to overlap. The formula, in a discrete time scenario like the one considered here, is:
    [0054] The value of the cross correlation is compared to a threshold, which can preferably be configured. The correlation can be normalized to energy by dividing Rxy by a norm factor, obtained, for example, as follows:
    K = £ x 2 (t> / t
    Ey = jy 2 (t) c / r Norm = VEx E y [0055] Figure 5 is a schematic flowchart of the wedding verification phase 220c, according to an embodiment of the present invention.
    [0056] Once the time window is identified (block 505) using the segmentation procedure described above, repetitions are performed between all classes (groups) of devices / devices (that is,
    Petition 870190087388, of 9/5/2019, p. 25/60 / 24 washing machine, refrigerator, dishwasher, oven, microwave oven, television set, personal computer, etc.) that could be recognized in the data flow received by sensor nodes 110. A current index i in the flowchart it is used to iterate the procedure for different classes of devices / devices; at each iteration, a new class of devices is selected (blocks 510, 515 and 517).
    [0057] For each class of devices / devices, one or more training sets, for example, a family or group of training sets, is available and is used for the wedding verification; each group of training sets contains a set of signal patterns to be used to attempt to match the data within the current time window. When more than one training set is available, the training sets (at least some of them, possibly all) within the group corresponding to the selected device / device group are considered in succession (current index j in the flowchart, and blocks 520 and 525 ).
    [0058] To assess the match between the selected training set j and the data within the current time window, the normalized cross correlation in energy is calculated as described above (block 530).
    [0059] If the normalized cross correlation in calculated energy is below a predetermined minimum threshold MINthreshold (block 535, output branch Y), that is, if the correlation between the selected training set and the data in the current time window is weak furthermore, this means that, with a probability proportional to (1 - MINthreshold), the data contained in the current time window does not correspond to any device / device belonging to the device group currently selected. No further attempts are made with other family training sets, and the next group of devices is then selected (block 517) to look for a potential match in the data set,
    Petition 870190087388, of 9/5/2019, p. 26/60 / 24 as long as there are additional device / device groups yet to be attempted (block 540 output branch Y).
    [0060] On the other hand, if the normalized cross-correlation in calculated energy is not below the minimum threshold MINthreshold (block 535, output branch N), it is checked whether the normalized cross-correlation in calculated energy is below or above a maximum threshold predetermined MAXthreshold (block 545). If the normalized cross correlation in calculated energy is below the MAXthreshold maximum threshold (block 545, output branch N), a new training set j = j + 1 is selected for the device / device group j under consideration (block 525 ), in order to iterate the pattern match and see if it is possible to improve the match probability by checking the cross-correlation against the new training set (block 555). If, however, there is no further training set available for the currently selected device / device group (block 560, outgoing extension N), the currently selected device / device group i is not placed in a device / device list. candidate responsible for data generation (block 565), and the procedure is iterated over a new device / device group (block 515), after evaluating whether other device / device groups exist (decision block 567, output Y - others device / device groups available - or output N - no device / device groups to be tested).
    [0061] In the case that the calculated cross correlation is greater than the maximum MAXthreshold threshold (block 545, output branch Y), then the currently selected class of devices / devices is put in the candidate list (block 570).
    [0062] At the end of all iterations, if the candidate list is empty, the currently selected time window is discarded.
    [0063] The process described above is repeated for several windows
    Petition 870190087388, of 9/5/2019, p. 27/60 / 24 time periods, for example, successive (minimum 1; the number of successive time windows depends on the number of event groups that someone wants to take into account).
    [0064] In this way, a list of candidate devices / devices (classes of) is constructed that contains candidate devices / devices (classes of) for which the cross correlation (normalized in energy) between the data measured by the sensors and the standards reference points defined by the training sets, in at least one of the time windows considered, is above the maximum threshold MAXthreshold.
    [0065] If there are candidates on the list, the device / device that generated the collected data flow is one of the devices / devices on the list of candidates.
    [0066] In accordance with an embodiment of the present invention, a cross comparison is performed, to reduce the likelihood of "false positives" included in the candidate list, and thus improving the reliability of the match (block 575).
    [0067] Reference is made to Figure 6. As mentioned above, the matching process is operated separately in each time window (index i, from 1 to R), in relation to different classes of devices / devices (training set - TS - index j, from 1 to N), each class having associated with it one or more (possibly a family of) training sets, created, for example, as described in the background.
    [0068] Each TS training set (or each family of training sets) corresponds to and identifies a single class of devices / appliances (eg washing machine, dishwasher, refrigerator, oven, etc.).
    [0069] For each class of devices / apparatus, an arrangement of R elements 6051 - 605N is created, each element of the arrangement corresponding to the result of the matching process (ie calculation of the cross correlation and
    Petition 870190087388, of 9/5/2019, p. 28/60 / 24 compared to the MINthreshold and MAXthreshold thresholds) in a respective of the R time windows. In the 6051 - 605n arrangements, the result of every match is indicated, for example, as a “1” in the event that the event is recognized (that is, the calculated cross correlation is above the MAXthreshold maximum threshold), and is indicated as “ 0 ”if the event is not recognized (that is, the calculated cross-correlation is below the maximum threshold MAXthreshold). In this way, N arrangements are created, containing the results of the matching process in every time window i of the set of R time windows.
    [0070] Then, the N arrangements are joined together to create a single matrix of 610 matching results, of RxN dimensions. In matrix 610, the generic row i provides the results of the matching process in the R temporal windows for a certain class of devices / devices, while the generic column i provides the results of the matching process in the generic time window in relation to the different classes of devices / appliances.
    [0071] The combined results matrix 610 can be scanned column by column, looking for those time windows in which an event was recognized and matched by more than one training set, that is, by more than one class of devices / devices. For example, if the scan for column i (i-th time window) returns a “0”, the event in the time window was not recognized using any training set, that is, none of the device / device classes matches the device / device that generated the data flow; if the matrix scan returns a single “1”, in row j (block 615, outgoing branch N), the event is recognized as having been caused by a device / device belonging to the class of devices / devices in row j (block 620). If the matrix scan returns more “1” s (block 615, output branch Y), meaning that more than one device / device class matched the event, a second comparison is performed using the correlation value
    Petition 870190087388, of 9/5/2019, p. 29/60 / 24 cross calculated in time window i. Then, the time window event i is recognized as caused by a device / device of the class for which the highest cross correlation was obtained (block 625).
    [0072] As an alternative to performing the selection based only on comparing the calculated value of the cross correlation, the cross correlation functions and their characteristic parameters such as average value, relative maximum values, standard deviation or others can also be taken into account and analyzed. In this way, an increased number of parameters can be used to improve the reliability of the result of the matching process.
    [0073] A different technique can also be applied with certain classes of devices / devices that generate pseudoperiodic data sets (such as refrigerators or boilers): in such cases, exploring an analysis in the frequency domain instead of the time domain can simplify the marriage process. In this case, in addition to the segmentation process described above that motivates the recognition of events, a wider time window containing several pseudo-periods of the signal is considered. Both the data set and the training sets are transformed into the frequency domain (using, for example, a Fourier transform). The transformed signals have the advantage to enhance the harmonics of pseudo-periodic behaviors leading to a faster recognition of the signals in the time domain; recognition can be performed using the pattern matching technique with the cross correlation described above.
    [0074] Possible alternative embodiments of the invention could be the use of a distributed algorithm to operate the signal classification: in this case the processing could be performed locally by the sensor nodes if the processing capabilities of the nodes allow this.
    [0075] The method according to the embodiment described here of
    Petition 870190087388, of 9/5/2019, p. The present invention can be performed as a software implementation at the network connection point 115 (typically, a device having limited computational resources) or as a remote middleware in the event that the data gathered from the sensor nodes is recorded in a remote database and processed remotely on a more powerful device.
    [0076] Alternatively or in combination, at least part of the method described previously can be implemented directly in the smart receptacles 110. Figure 7 shows schematically the main components of an intelligent receptacle 110: block 705 denotes one or more sensors of electrical quantities, as current, voltage or power, for example, a current meter or a voltage meter. The sensor 705 sends the measured data to a microprocessor or microcontroller 710. A transceiver 715 enables, for example, wireless communications with connection point 115 and other nodes 110 on the network. The method according to the present invention can be performed by the microprocessor 710 which executes a specific program.
    [0077] The described method can also be used to automatically detect events on the monitored devices / devices, such as, for example, changes in the devices' own behavior based on which imminent failures or degradation of their performance can be predicted.
    [0078] The present invention is applicable and useful in different technical fields such as home energy management system in order to detect devices connected to the main power network using a home area network such as a WSN to operate energy management applications. energy (for example, readiness control, peak impediment), and commercial energy management systems, in order to detect an event related to energy monitoring and operate self-awareness of monitored devices using WSN.
    Petition 870190087388, of 9/5/2019, p. 31/60 / 24 [0079] The present invention has been set out here describing some embodiments of it. Those skilled in the art will be able to devise various modifications to the described embodiments, without departing from the extent of protection defined in the attached claims. For example, other types of home area networks can be used instead of WSNs to network sensor nodes, such as Power Line Carrier (PLC), and more generally ultra low power networks (for example, Bluetooth, WiFi ), or also a wired LAN.
    [0080] Also, although the present invention has been described here with reference to electrical devices / appliances, the present invention can also be applied to non-electrical devices / appliances, such as thermal energy devices / appliances, for example, a system of independent heat radiators, each equipped with its own thermostat, and thus each exhibiting a peculiar behavior in terms of absorbed thermal energy. In this case, a sensor can be provided in operative association with each device / device that transduces the monitored quantity, for example, thermal energy, in an electrical signal, and the analysis can thus be performed on the transduced electrical signal.
    Petition 870190087388, of 9/5/2019, p. 32/60
    权利要求:
    Claims (13)
    [1]
    1. System for the automatic identification of a device or apparatus (105a-105c), characterized by the fact that the system comprises:
    - at least one sensor (110) configured to be associated with the device or device to be identified, and to monitor a time evolution of at least an electrical amount indicative of a power consumption by the device or device;
    - an analyzer (110; 115; 125) in communication relation with said at least one sensor and configured to receive reports from it (215) of the monitored electrical quantity, the analyzer being furthermore configured to automatically identify the device or apparatus being analyzed (220) the time evolution of the monitored electrical quantity, in which said analysis comprises:
    - calculate a cross correlation (530, 555) between the evolution in time of the monitored electrical quantity and at least one reference standard representative of at least one device or sample device;
    - insert at least one device or sample device in a list of candidate devices or devices in the event that a value related to the calculated cross-correlation is above a predetermined threshold;
    - in the event that more than one sample device or device is included in the candidate list, identify the device or device by performing a selection between the candidate devices or devices based on characteristic parameters related to the respective and different calculated correlations of said value.
    [2]
    2. System according to claim 1, characterized by the fact that said at least one reference standard includes at least two
    Petition 870190087388, of 9/5/2019, p. 33/60
    2/4 reference standards, each corresponding to a respective observed evolution in time of the electrical quantity monitored in relation to a specific operational mode of the same device or sample device.
    [3]
    3. System according to claim 1, characterized by the fact that said at least one reference standard is generated by calculating, on average, the observed evolutions in time of the monitored electrical quantities of devices or devices.
    [4]
    4. System according to claim 1, characterized by the fact that said at least one reference standard is generated by observing the evolution in time of the monitored electrical quantity of the device or device to be identified during a first operating cycle of it.
    [5]
    5. System according to claim 4, characterized by the fact that at least one reference pattern is generated by detecting characteristic events in the time observed in the monitored electrical quantity.
    [6]
    6. System according to claim 5, characterized by the fact that detecting said characteristic events includes detecting peaks in the time evolution of the monitored electrical quantity.
    [7]
    7. System according to claim 1, characterized by the fact that calculating this cross-correlation includes determining a time observation window and cross-correlating the monitored electrical quantity and at least one reference pattern in at least one time corresponding to the determined observation window.
    [8]
    8. System according to claim 7, characterized by the fact that calculating said cross-correlation is repeated in at least two different time intervals of duration equal to the determined time window.
    [9]
    9. System according to claim 7, characterized by the
    Petition 870190087388, of 9/5/2019, p. 34/60
    3/4 the fact that said time window of observation is determined by detecting characteristic events in the observed evolution in time of the monitored electrical quantity.
    [10]
    10. System according to any one of claims 1 to 9, characterized by the fact that said analysis includes, before calculating the cross correlation, transforming the observed evolution in time of the monitored electrical quantity and at least one reference standard in the domain of frequency, and in which to calculate this cross-correlation is performed in the results of the transformation.
    [11]
    System according to any one of claims 1 to 10, characterized in that said characteristic parameters include at least one between an average value, relative maximum values, standard deviation.
    [12]
    System according to any one of claims 1 to 11, characterized in that the device or apparatus is an electrical device or apparatus, and in which the system is embedded in an intelligent receptacle (110), whereby the device or electrical appliance can be connected to an electricity receptacle.
    [13]
    13. Method for the automatic identification of a device or apparatus (105a-105c), characterized by the fact that the method comprises:
    - associate at least one sensor (110) with the device or device to be identified, the sensor being configured to monitor a time evolution of at least an electrical amount indicative of energy consumption by the device or device;
    - receive data on the monitored electrical quantity informed by at least one sensor, and automatically identify the device or apparatus analyzing the evolution in time of the monitored electrical quantity, in which said analysis comprises:
    Petition 870190087388, of 9/5/2019, p. 35/60
    4/4
    - calculate a cross correlation (530, 555) between the evolution in time of the monitored electrical quantity and at least one reference standard representative of at least one device or sample device;
    - insert at least one device or sample device in a list of candidate devices or devices in the event that a value related to the calculated cross-correlation is above a predetermined threshold;
    in the event that more than one sample device or device is included in the candidate list, identify the device or device by performing a selection among the candidate devices or devices based on characteristic parameters related to the respective and different calculated correlations of said value.
    类似技术:
    公开号 | 公开日 | 专利标题
    BR112013002490B1|2020-01-28|system and method for the automatic identification of a device or apparatus
    Kim et al.2009|ViridiScope: design and implementation of a fine grained power monitoring system for homes
    Kleiminger et al.2013|Occupancy detection from electricity consumption data
    Schoofs et al.2010|Annot: Automated electricity data annotation using wireless sensor networks
    ES2642210T3|2017-11-15|Monitoring system
    Choi et al.2018|Detecting and identifying faulty IoT devices in smart home with context extraction
    KR101986838B1|2019-06-07|SAFETY MANAGEMENT SYSTEM FOR MEASUREMENT DEEP-LEARNING AND CONTROL SENSOR BASED LoRa NETWORK
    US20180034657A1|2018-02-01|Method and System of Monitoring Appliance Usage
    KR20140116619A|2014-10-06|System and method for predicting energy consumption
    CN106773763B|2019-09-06|A kind of information linkage processing method and device
    KR101872813B1|2018-06-29|Method and apparatus for determining abnormal events related to in-home person
    CN105485831A|2016-04-13|Method and apparatus for managing heating, ventilation, and air conditioning
    WO2017066435A1|2017-04-20|Sensor data analytics and alarm management
    US20160049789A1|2016-02-18|Energy management system
    Ding et al.2021|IOTSAFE: Enforcing Safety and Security Policy with Real IoT Physical Interaction Discovery
    WO2020035696A1|2020-02-20|Sensing fluid flow for estimating fluid flow state
    Reinhardt et al.2014|Can smart plugs predict electric power consumption? A case study
    WO2015041577A1|2015-03-26|Methods, apparatuses, and computer program products for configuring and collecting information from sleepy devices
    Abreu et al.2014|QoS-based management of biomedical wireless sensor networks for patient monitoring
    Molina et al.2014|Model based design of distributed embedded cyber physical systems
    Jafari et al.2017|A testbed evaluation of MAC layer protocols for smart home remote monitoring of the elderly mobility pattern
    KR101457028B1|2014-10-31|System and method for remote inspecting
    KR20180012038A|2018-02-05|Method for managing electronic equipment using internet of things service and apparatus for the same
    Gil et al.2018|Data anomaly detection in wireless sensor networks with application to an oil refinery
    Padin et al.2017|Diagnosing False Data Injection Attacks in the Smart Grid: a Practical Framework for Home-Area Networks
    同族专利:
    公开号 | 公开日
    WO2012016589A1|2012-02-09|
    BR112013002490A2|2016-05-31|
    EP2601482B1|2016-01-06|
    US20130132008A1|2013-05-23|
    US9606153B2|2017-03-28|
    EP2601482A1|2013-06-12|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    US4858141A|1986-04-14|1989-08-15|Massachusetts Institute Of Technology|Non-intrusive appliance monitor apparatus|
    US5717325A|1994-03-24|1998-02-10|Massachusetts Institute Of Technology|Multiprocessing transient event detector for use in a nonintrusive electrical load monitoring system|
    WO2007139587A1|2006-05-26|2007-12-06|The Board Of Regents Of The Nevada System Of Higher Education On Behalf Of The Desert Research Institute|Utility monitoring systems and methods of use|
    GB0713515D0|2007-07-12|2007-08-22|Ampy Metering Ltd|SMS smart credit metering|
    US8255090B2|2008-02-01|2012-08-28|Energyhub|System and method for home energy monitor and control|
    GB0803140D0|2008-02-21|2008-03-26|Sentec Ltd|Technique for inference of multiple appliances' power use from single point measurements|
    WO2010053562A2|2008-11-06|2010-05-14|Silver Springs Networks, Inc.|System and method for identifying power usage issues|GB2488164A|2011-02-18|2012-08-22|Globosense Ltd|Identifying electrical appliances and their power consumption from energy data|
    JP5546624B2|2011-12-09|2014-07-09|三菱重工業株式会社|Windmill wing|
    US9198204B2|2012-04-11|2015-11-24|Google Inc.|Apparatus and method for seamless commissioning of wireless devices|
    US10075334B1|2012-04-11|2018-09-11|Google Llc|Systems and methods for commissioning a smart hub device|
    BE1020861A4|2012-08-13|2014-06-03|Fifthplay N V|METHOD AND APPARATUS FOR MEASURING THE CONSUMPTION OF AN ELECTRIC APPLIANCE.|
    US9294246B2|2013-03-19|2016-03-22|Electronics And Telecommunications Research Institute|Wireless communication device using common control channel and wireless communication method using the same|
    US9294469B2|2013-07-03|2016-03-22|General Electric Company|Systems and methods for establishing a connection between an appliance and a home energy management device|
    TWI517079B|2013-07-30|2016-01-11|財團法人工業技術研究院|Method for identifying electronic device and apparatus and system using the same|
    KR102125331B1|2013-11-07|2020-06-22|삼성전자 주식회사|Method and apparatus for detecting device connected to a smart plug|
    KR20150054123A|2013-11-11|2015-05-20|삼성전자주식회사|Electronic Device And Method for Detecting External Device Of The Same|
    TWI497861B|2014-01-22|2015-08-21|Chunghwa Telecom Co Ltd|Cloud control system and cloud control method thereof|
    KR102132245B1|2014-03-19|2020-07-09|삼성전자 주식회사|Method for automatic mapping between smart plug and electronic device in a home gateway and apparatus thereof|
    CN104111641B|2014-05-30|2017-07-28|小米科技有限责任公司|Power consumption statistical method, device and system|
    US20160197998A1|2015-01-02|2016-07-07|Xeros Limited|Monitoring system|
    WO2016108051A1|2015-01-02|2016-07-07|Xeros Limited|Monitoring system|
    US20160195409A1|2015-01-02|2016-07-07|Xeros Limited|Monitoring system|
    US10489715B2|2015-03-27|2019-11-26|International Business Machines Corporation|Fingerprinting and matching log streams|
    US10489714B2|2015-03-27|2019-11-26|International Business Machines Corporation|Fingerprinting and matching log streams|
    CN105467874A|2015-11-30|2016-04-06|小米科技有限责任公司|Method and device for identifying categories of electronic devices on intelligent sockets|
    US9973521B2|2015-12-28|2018-05-15|International Business Machines Corporation|System and method for field extraction of data contained within a log stream|
    DE102016111351A1|2016-06-21|2017-12-21|Emh Metering Gmbh & Co. Kg|Electric household appliance with a data transmission device and device for receiving data from such a device|
    US10613493B2|2016-08-03|2020-04-07|Chang Liu|Fusion sensor wireless decision device and system for using the same|
    IT201700121670A1|2017-10-26|2019-04-26|Telecom Italia Spa|System and method to manage electricity supply through certified measures|
    US11126244B1|2018-09-10|2021-09-21|Amazon Technologies, Inc.|Power line presence detection|
    US11100458B2|2018-09-14|2021-08-24|International Business Machines Corporation|Asset and device management|
    DE102018222376A1|2018-12-19|2020-06-25|Robert Bosch Gmbh|Method and device for adapting a functionality of a sensor of a sensor network|
    法律状态:
    2019-01-15| B06F| Objections, documents and/or translations needed after an examination request according [chapter 6.6 patent gazette]|
    2019-07-09| B06T| Formal requirements before examination [chapter 6.20 patent gazette]|
    2019-12-03| B09A| Decision: intention to grant [chapter 9.1 patent gazette]|
    2020-01-28| B16A| Patent or certificate of addition of invention granted|Free format text: PRAZO DE VALIDADE: 20 (VINTE) ANOS CONTADOS A PARTIR DE 04/08/2010, OBSERVADAS AS CONDICOES LEGAIS. |
    优先权:
    申请号 | 申请日 | 专利标题
    PCT/EP2010/061341|WO2012016589A1|2010-08-04|2010-08-04|System and method for the automatic identification of electric devices/appliances|
    [返回顶部]