• 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:
    The invention concerns a method for reliable detecting and monitoring of a multitude of different earth faults in three-phase electric power distribution networks. A method is presented which combines a transient estimation method with a modified phasor based wattmetric method. The transient method estimates the capacitance and conductance of each feeder and also for the neutral point device, and uses this to find the faulted feeder. The modification of the wattmetric method is to use the negative derivative of the neutral point voltage phasor magnitude to identify when the fault is not connected and then disable (block) the active power criterion.
    公开号:SE1150535A1
    申请号:SE1150535
    申请日:2011-06-14
    公开日:2012-12-15
    发明作者:Magnus Akke
    申请人:Dlaboratory Sweden Ab；
    IPC主号:
    专利说明:

    2 coupling to minimize safety risks and damage to equipment. Direct grounding of distribution networks is mainly used in the USA.
    An alternative is that the zero point is isolated from earth. In this case, the fault current is returned via the capacitive coupling between healthy phase conductors and earth.
    Relatively little fault current flows through this capacitive connection and therefore it is more difficult to detect earth faults than with a directly earthed network. isolated zero point is mainly used in smaller overhead line networks, without cables. In another type of system earthing, the zero point is connected to earth via an impedance, which in comparison with direct earthing, reduces the fault current and the risk of damage. System grounding that gives low fault currents is usually combined in the category, high-ohmic earthed networks. Networks with isolated zero point fall into this category.
    High-density earthed networks usually provide better operating characteristics. If the current through the fault point can be reduced to a very low value, then maintained electricity supply is made possible in the event of a ground fault. A common method of reducing the current through the fault point is to use a coil at the zero point of the feeding station. The coil compensates it out or diverts the capacitive fault current.
    High-impedance earth faults are typically caused by fallen phase conductors, or by accidents where people or animals have come into contact with live system parts, which poses a danger to life. It is desirable to be able to safely detect high-impedance earth faults, which is usually referred to as "the sensitivity of earth fault protection".
    There are also factors that counteract the sensitivity of the protection. During normal operation, there is always a small current that goes to earth, the size of which depends on the natural asymmetry in the network's capacitance ground. In any case, earthing with the Petersen coil has an overall positive impact on the network's operational reliability and quality, especially for overhead line networks.
    Protection based on power direction measures the sum current for an output and multiplies by the zero point voltage. The product becomes proportional to active power and is therefore called wattmetric earth-fault protection. The power direction can be used to point out the faulty output. Relay protection is often used based on zero point voltage as reserve protection for earth faults in the mains. Oh PCSSQSENNIHIILGW IMZ-OÅ-lßver B 10 15 20 25 30 3 If the wattmetric earth-fault protection does not work, the backup protection based on the zero point voltage will disconnect all outputs in the transformer station.
    Grounding with a Petersen coil means that the mains, during earth faults, are stressed more by high voltage and can therefore lead to double earth faults. The first earth fault increases the voltage on the healthy phases, which increases the risk of a second earth fault in another phase. This is called a double earth fault, or in English “cross-country” fault, and the high currents that occur can have more serious consequences than a single ground fault.
    Extinguishing an arc with the help of a Petersen coil is often used in medium voltage networks. Figure 1 illustrates a Petersen-grounded distribution network with three outputs, one of which is faulty and two are healthy. Figure 2 illustrates a simple equivalent circuit, which can be used to describe the occurrence of the zero point voltage, and the relationship between the sum current of the outputs and the current through the zero point equipment.
    In any case, there are significant difficulties for networks with Petersenspole; the biggest difficulty is to properly handle all the types of earth faults that occur. More specifically, "handle" means both the correct function of conventional relay protection, but also to provide correct information about all types of disturbances where the phase conductor insulation against earth deviates from normal mains operation. So the term “handle” is used as a much broader term that intends to correctly report information on all types of earth faults that give deviations from normal mains operation, in either zero point voltage or in the sum current at some output.
    DISADVANTAGES OF PREVIOUS PATENT A conventional earth-fault protection is described in document DEA4413068. Conventional earth-fault protection and detectors are designed for permanent earth-fault. Conventional detection methods are mainly based on phase vectors and Fourier calculations. These methods aim at stationary sinusoidal signals. Therefore, the results from these methods may be poor or even incorrect for disturbances of shorter duration than half a fundamental tone cycle. It is a fact that the transient itself contains abundant Ah PßSQSEOO pl lv fl nlldøc 2012M13 vøf. 9 10 15 20 25 30 4 with information that can be used to advantage for error detection, but fundamental tone-based Fourier transform actually filters out the high-frequency components. Therefore, conventional methods based on phase vectors are less suitable for processing transient and non-sinusoidal signals.
    The conventional earth-fault protection and detectors therefore work randomly in connection with other types of earth-fault and provide uncertain information.
    EP0999633 describes a method based on phase vectors which stores the signal before and after errors, and uses this to distinguish the faulty output. The methodology of subtracting the signal before the fault, from the signal after the fault, is often called "delta quantities" by relay protection engineers. The main disadvantage of the method is the need to define a criterion for separating the signal before error from the signal after error. The method works for well-defined transitions from a healthy network to a faulty network. However, the said method becomes unreliable and less useful for changing errors, where the network goes through a sequence of different error conditions and error types. Another disadvantage is that the method uses information from only the fundamental tone, which limits its usability for transient disturbances.
    There are methods that are specifically designed for re-igniting (intermittent) earth faults, which often have recurring transients in voltage and current. An example is described in WO9912048A1. The basic idea of WO9912048A1 is to compare the polarity of the current spikes with each other, or with the sign of the instantaneous value of faulty phase voltage. The disadvantage of the method is that only single measured values (samples) are used for decision making, which makes it unreliable. The method can possibly work for a limited number of errors, where the error mode is several kilometers out on the output, so that the signal does not contain any transients with a high frequency.
    As illustrated by the interference detection in Figure 7B, it becomes almost impossible to use individual samples for decision making when the ground fault gives rise to high frequency signals. Figure 7B is representative of a re-igniting earth fault near the supply transformer station, because it is doubtful whether the method iWO9912048A1 gives reliable results for nearby earth faults. It is also doubtful whether the concept of calculating current peaks is to- An NSSQSEOO ps of fl nal fi oc 2000443 var. H 10 15 20 25 30 5 reliable for high-frequency current signals which may consist of the sum of wave propagation and various reflections in the network.
    For Petersen-grounded networks, for example as described in WO2010115474A1, active power is used to determine faulty output. The innovation in WO2010115474A1 is that the power calculation is based on true RMS value, instead of conventional phase vectors that calculate the active power for only the fundamental tone.
    As previously mentioned, conventional phase vectors are less useful for transient signals. However, they can still be useful for signals where no clear transient can be identified. Conventional relay protection, based on phase vectors and active power, has problems functioning during transient / intermittent situations. A method is needed that solves this.
    Therefore, the invention focuses in the application on being able to detect a variety of different types of earth faults. A directional method is presented that uses transient information to distinguish the faulty output from the healthy outputs. The method presented in the next section offers a solution, by combining a transient method with a modified phase vector method based on active power (wattmetric), SUMMARY OF THE INVENTION The invention relates to a method and equipment for a reliable detection method and monitoring of a variety of earth faults. in a three-phase electricity grid for electricity distribution. A method is presented that combines a transient method for estimation, with a modified phase vector method based on active power (wattmetric). The transient method estimates the capacitance and conductivity of each output and also of the zero point equipment, and this is used to determine the faulty output. The modification of the wattmetric method consists of using the negative derivative of the amount of the zero point voltage, to determine when the fault is not connected and then deactivate the criterion that uses active power.
    Ah PßSQSEDO p! by flnllAøc 2012-04-13 v01. BRIEF DESCRIPTION OF THE DRAWINGS In order to make the manner in which the above and other advantages and objects of the invention easier to understand, a more detailed description of the above invention will be given with reference to certain embodiments. which are shown in the following drawings.
    Noting that these drawings show only typical embodiments of the invention and therefore should not be construed as limiting its scope, the invention will be described and explained in detail and with further details with reference to the accompanying drawings, in which: Figure 1 schematically shows a system for electricity distribution.
    Figure 2 schematically illustrates the relationship between the zero point current and the sum currents of the outputs.
    Figure 3A shows a graph with the zero point voltage and the summed currents over a time period of 500 ms, where a transient earth fault occurs.
    Figure 3B is a close-up of Figure 3A.
    Figure 4A shows a graph with zero point voltage and sum current during a time period of 1 second, where a temporary earth fault with fundamental frequency occurs.
    Figure 4B is a close-up of Figure 4A.
    Figure 5A shows a graph with zero point voltage and sum current during a time period of 10 seconds, where an intermittent earth fault with 5 strokes occurs.
    Figure 5B is a close-up of Figure 5A.
    Figure 6A shows a graph with zero point voltage and sum current during a time period of 1 second, where an intermittent earth fault with 6 strokes occurs.
    Figure 6B is a close-up of Figure 6A.
    Figure 7A shows a graph with zero point voltage and sum current during a time period of 500 ms, where an intermittent earth fault of 9 strokes occurs.
    Figure 7B is a close-up of Figure 7A.
    Ah P4359 $ E00 p! tvllnllßoc 2012-04-13 v19 10 15 20 25 30 7 Figure 8A shows a graph with phase voltages, zero point voltage and sum current during a time period of 800 ms, where a variable error occurs.
    Figure 8B is a close-up of Figure 8A.
    Figure 9 shows a flow chart for the combined method according to the present description.
    Figure 10 illustrates an implementation of the transient method according to the present description.
    Figure 11 shows a flow chart of the modified wattmetric method according to the present description.
    Figure 12A shows a graph with three phase voltages, zero point voltage and sum current over a time period of 400 ms, where a transient earth fault occurs.
    Figure 12B shows a graph with a close-up of the sum current during a time period of 30 ms, where a transient earth fault occurs.
    Figure 12C shows a graph with a close-up of the estimated conductance G and the capacitance C, during a time period of 30 ms where a transient earth fault occurs.
    Figure 13A shows a graph of zero point voltage and sum current, over a time period of 800 ms, where an intermittent earth fault occurs.
    Figure 13B shows a graph with a close-up of the sum current during a time period of 80 ms, where an intermittent earth fault occurs.
    Figure 13C shows a graph with close-up of estimated conductance G and capacitance C, during a time period of 80 ms, where an intermittent earth fault occurs.
    Figure 13D shows a graph with a close-up of calculated phase vector for the amount of the zero point voltage, its derivatives and active power, during a time period of 80 ms, where an intermittent earth fault occurs.
    Ah P4359SE00 ps iv Ilnllßoc N12-04-13 vel. 9 10 15 20 25 30 8 DETAILED DESCRIPTION DIFFERENT TYPES OF GROUND ERRORS Several different types of earth faults occur in an electrical distribution network.
    These faults have very different properties, and the most important challenge for a pre-earth fault protection system is to be safe, reliable and sensitive to all types of earth faults.
    Therefore, this section exemplifies some different types of earth faults that have been recorded by a high-performance disturbance recorder from a (130 kV / 20 kV) transformer station in southern Sweden. The interference recorder was based on LABVIEW software and CRIO hardware (National Instrument, Austin, Texas, USA) with 50 kHz simultaneous sampling, 22 kHz effective bandwidth, 24-bit resolution and up to 10 seconds of recording length. Figures are used to illustrate the different types of earth faults.
    REMAINING EARTH FAULT Some of the faults establish a relatively stable connection between a phase conductor and earth. For a permanent fault in a cable network, it is required that the faulty cable part is disconnected from the network. For overhead line networks, it is common for earth faults to have an arc in the fault location. For example, bird droppings can contaminate the surface of an air insulator, which can ignite an arc. An advantage of the Petersen coil is that it helps to reduce the fault current, so these faults often disappear by themselves, which is usually called "self-extinguishing earth faults". In this case, no relay function is needed. Therefore, the trip signal from the relay protection is often delayed by a few seconds, which gives time for the fault to self-extinguish. If the error time is longer, the arc can be eliminated by disconnecting, that is, making the wire de-energized for a few seconds. Regardless of whether the permanent faults have an electric arc, or not, they can be easily identified and eliminated using generally known relay protection technology.
    Ah PÅCSQSEOOM lv flnildøc 2012-04-131 / 07. 9 10 15 20 25 30 TRANSIENT EARTH FAULTS Figures 3A and 3B illustrate a ground fault with a single transient. It can be assumed that this type of fault can result from insulation penetration caused by moisture, which gives an arc that evaporates the moisture, and then self-heals the insulation and quickly disconnects the fault.
    The lower graph in Figure 3A shows the sum of the phase current, also called zero-current current (actually three times the zero-current current) and shows a short-lived transient with a high amplitude, almost 1000 A. The upper graph in Figure 3A shows the zero point voltage that oscillates due to resonance between the Petersen coil inductance and the capacitance of the outputs to ground. Phenomenon can be considered as the sum current giving an impulse which excites the resonant circuit consisting of the Petersen coil parallel to the capacitance of the outputs, as depicted in Figure 2. Figure 3B shows a close-up of Figure 3A and shows the transient when the fault is connected.
    It is important to understand that the fault is only connected for a very short period of time. When the fault is switched on, the resonant circuit is disturbed from its equilibrium. After the fault has been disconnected, the mains is healthy, but since the resonant circuit is not in equilibrium, it will continue to oscillate with decreasing amplitude going towards zero. Therefore, the amount of the zero point voltage is in itself an insufficient criterion for a connected earth fault. Information about the derivatives of the zero point voltage is also needed to obtain a reliable criterion for a connected earth fault.
    For a single transient, as shown in Figure 3A, it is not necessary to trip the switch and disconnect the output. However, the transient contains plenty of information that should be used. The fault should be reported and the incorrect output should be identified. The information is useful for preventive maintenance and also as an early warning system. In summary, it is important that a method for earth faults can also handle this type of earth fault.
    TEMPORARY SOIL ERRORS WITH BASIC TONE FREQUENCY Figures 4A and 4B show temporary earth faults that heal on their own and disappear after about 50 ms. In contrast to transient earth faults, both current and voltage are- Oh PGSSSEOO p! tv flnalßøc 2012-04-13 vil, 9 10 15 20 25 30 10 the amount is small, and the signals are mainly of fundamental frequency.
    Therefore, the signal contains some transient information with short duration (less than 20 ms) that can be used for analysis. Therefore, this type of ground fault should be analyzed with methods based on the fundamental tone, such as DFT phase vectors.
    INTERMITTENTA (RETURNING) EARTH FAULT This type of earth fault is called intermittent, or re-igniting, earth fault.
    Contact with earth, occurs in a sequence where the fault re-ignites and disappears. This process can be repeated several times. If not the faulty wiring section, disconnected from the network, the process causes a long-term disturbance in the mains. Re-igniting earth faults are common in Petersen-earthed networks and often occur in connection with damaged insulation. Intermittent faults can occur in places where the cable insulation, connections, or joints are damaged due to mechanical stress, material defects, or aging. A re-igniting earth fault can also occur in a newly buried cable as a result of mechanical stress caused during the relocation work.
    Three different types of intermittent earth faults are presented which clearly show that intermittent faults can have very different properties.
    Figure 5A and the close-up of 5B show the first example of an intermittent ground fault with five strokes within 10 seconds of recording time. The characteristic of the disturbance is that the time between each stroke is so long that all quantities fall to zero, before the next stroke occurs. Therefore, there is no overlap between the strokes and each new stroke arises from a state of equilibrium, where all quantities are close to zero. For this type of intermittent fault, the zero point voltage drops almost to zero, which means that timers will be reset to fault-free condition and the risk of unwanted. tripping via the time-delayed zero-point voltage protection is eliminated.
    Figure 6A and the close-up of 6B show the second example of an intermittent ground fault with five strokes in less than one second of recording time. The characteristic of the disturbance is that the time between each stroke is so short that the voltage transient only subsides to around half, before the next stroke takes place. There- Ah P435 $ SE00 ps lv fi nlldoc 2012M13 v61. 9 10 15 20 25 30 11 for overlapping transients from the different kinds of each other. For this type of intermittent fault, the zero point voltage is still high, as long as the intermittent fault is present. If the duration of intermittent faults exceeds the time delay for the zero point voltage protection, there is an obvious risk of unwanted disconnection. This is bad and undesirable, as this protection is a backup protection and disconnects all outputs in the entire network station.
    Figure 7A and the close-up of 7B show the third example of an intermittent ground fault with a plurality of high frequency current peaks in less than one second of recording time. The characteristic feature of the disturbance is the effect of the high frequency content of the sum current. The reason is that the fault condition is very close to the point where the disturbance has been recorded. As can be seen from the close-up in Figure 7B, the sum current for the outputs has several peaks with opposite signs even within a time frame of less than 1 millisecond.
    TWO Simultaneous GROUND ERRORS IN DIFFERENT SITES AND PHASES (CROSS COUNTRY) In a mains ground with Petersen coil, the first earth fault causes an increase in the voltage on the healthy phases. This is illustrated in Figure 8A, during the time period from 200 ms to 400 ms. The increase in voltage on the healthy phases increases voltage stress on other parts of the network. If any part of the healthy network contains a component, or plant part, starting with insulation penetration, the risk increases that a second earth fault will occur. This is shown in the lower graph in Figure 8B, where the measuring range is at 600 A and 1000 A, respectively. Two earth faults in different phases will almost function as a short circuit, which gives a very large current. The type of fault is called a "cross-country" fault and can have much more serious consequences than a simple ground fault.
    CHANGING EARTH ERRORS (EVOLVING) Some methods for detecting earth errors are based on the assumption that the network goes from a well-defined healthy state to an incorrect state. Emeller- Oh PÅSSQSEN pl lv flnaldoc 2012-04-13 v01. 9 10 15 20 25 30 12 time, in many situations the errors occur in a sequence of several events, as shown in Figure 8A.
    Figure 8A shows the following sequence of events: 1) The first event is a single earth fault at the time of 295 ms; 2) The second event occurs at 464 ms and is a second earth fault but in a different phase, which results in a "cross country" fault (double earth fault in different places and phases) 3) The third event occurs at 550 ms and is the third failure of the last healthy phase, resulting in low zero voltage, but still strong currents.4) The fourth event occurs at 670 ms, when an output is disconnected due to overcurrent, and disconnects two of the faulty phases, leaving the network with a simple earth fault 5) The fifth event occurs at 680 ms, when the remaining earth fault self-extinguishes.
    As shown in Figure 9, the invention is a method which aims to detect and determine the direction of all types of earth faults. The invention combines two new methods, namely a transient method that estimates the conductivity and capacitance, and a fundamental-frequency method, called "active power (wattmetric) with blocking". Therefore, it is possible to handle a variety of different error types.
    The expression "sum current" is the same as "the sum of the phase currents" or three times the zero-sequence current (3 * 10). The motive for using a combined method can be explained by Figure 2. It shows that the sum current to the faulty output is the current through the zero point equipment and the sum of the sum current from all healthy outputs, but with the reverse sign, that is, negative sign. The inductance of the coil at the zero point means that the zero point current has a much slower dynamic than the currents of the outputs. This is also illustrated by the recording, for example Figure 3B.
    Figure 2 can be used to model the electrical properties of each output. These properties are modeled by two parameters, namely conductance, with the designation G = 1 / R, and capacitance, with the designation Oh P4359SE00 ps nv fl rlnldnc 20120013 v61. 9 10 15 20 25 30 13 the output C. For the healthy outputs, G and C will be estimated for each output. For incorrect output, the current has a negative sign. For fast transients, the current entering the output is approximately the sum of sum currents from the healthy feeds, but with the opposite sign. Thus, for fast transients, the estimation for the faulty output will give the sum of G and C for the healthy mats, but with the opposite sign, ie negative values.
    However, for slower dynamics, typically in the range of the fundamental frequency, the amount of coil current increases and can no longer be neglected. At 50 Hz, the coil should be set so that its current compensates for the capacitive currents of the mains. Therefore, the inductive coil current is in the same magnitude as the sum of capacitive currents from the outputs.
    If the coil current is slightly too large (overcompensated), this can result in the 50 Hz current entering the faulty output becoming inductive. On the other hand, if the coil current is slightly too small (undercompensated), then the current entering the faulty output may become capacitive. At 50 Hz, a positive inductive current looks the same as a negative capacitive current. Therefore, it is almost impossible to find the faulty output by using the relationship between the reactive and capacitive current components at 50 Hz. This is the reason why watt metric methods (active power) must be used at 50 Hz.
    The basic idea of the transient method is to estimate the two parameters G and C in the transient state. This idea works under transient conditions, but fails when the signal is dominated by fundamental frequency, that is, 50 Hz. The key is to use the current through the inductive zero point as a criterion when the method provides reliable estimates. So in addition to estimating G and C for all outputs, the two parameters are also calculated for the zero current. By doing so, it is possible to separate the time window where transients dominate from the time period where the signal is dominated by 50 Hz.
    The next section describes the transient method, which is assumed to be a new concept and therefore this part is extensive with more details. Then the method for the fundamental frequency is described, which is a modification Ah P4359SE00 p! W Ilnàlßøc 2012-04-13 vIf. 9 10 15 20 25 30 14 of a well-established idea to use active effect. Here is the new idea to use the derivative of the amount of the DFT phase vector for zero point voltage to identify when the fault is connected or not. This section is slightly shorter.
    The proposed method has been evaluated with interference registrations from a transformer station in southern Sweden. A high-performance disturbance detector has been active for 9 months and recorded more than 500 disturbances, of which approximately 70 have been of greater amplitude. The last part shows two examples of test results that use these recordings. An example is a single transient for which the transient method works perfectly. The second example is a disturbance with an intermittent fault, where there are several transients for a fraction of a second. The second example is used to illustrate when the method based on fundamental frequency may be useful, as an alternative when the transient method has insufficient information due to a large 50 Hz component.
    DESCRIPTION OF THE TRANSIENT METHOD The term "sum current" means "the sum of the phase currents", and "three times the zero sequence current (3 * | 0)".
    The sum current flowing into each output can be modeled as, f = Gu + Cïï m) m where u is the zero point voltage, where G and C are the conductance and capacitance of the output, respectively. Note that the equation is dual to the differential equation algorithm often used for distance protection, which is, u = Ri + L%. lnför the designation _Éi w_w. Qæ For a discrete system with the sampling interval, h, there are many methods to approximately calculate the voltage derivatives. A simple method is to use the time-discrete approximation Oh PßæSEw pl lv flnllßøc 2012-04-13 vel. 9 15 Wk “% (Uk'Uk-1) = LI-q Jak (zb) where q is the shift operator. For a finite set with n samples, the relation can be written I = UG (3) 5 where 'i ”1 Wi 1 = i U = u w km (4)::: C iuw The unknown parameters can be solved from G _ T -1 T [CJ _ (UU) U 1 (5) Calculations give 10 (UTUyI = w aayruTu) (e) where n fl n 2 data / TU) = [z 14,2) - [z wa - [Z ukwk] (7) k = lk = lk = l och fl n Wi Zukwk aamUTU) = § = 1 ti, zukwk Z ”i k = 1 k = 1 15 Based on its relations, G and C can be estimated as: Å 1 n n. nn 'G i-ídet (UTU) wš ukzkJ -Lš ukwk wkzkn (98) A 1 n 2 n I nn IC = Kšuk jíšwkzkj- [šukwkjíë ukzk fl (9b) (3) Åh PÅBSQSEOO pl iv fl nilßoc 2012-04-13 vIf . i 16 BEFORE COMPACT NAMES Use a n sampling window. Introduce the designations of the variance of the voltage and its derivatives (ÉufJ = var (u) = A (Éw, f] = var (w) = B (10) k = lk = l 5 Use the following designation for covariance between voltage and its derivata (Éukwk) = cov (u, w) = D (11) k = l Use the designation H for the determinant, then get H = AB-D2 (12) 10 introduce the following designations for covariance between the sum stream for output m and, voltage and its derivatives, respectively. [Éukim, kJ = cov (u, ím) = Em (13a) k = l (Éwkim, k¶ = cov (w, im) = Fm (13b) k = l With these designations, the estimate for output becomes m, 15 Gm = (143) H Öm = (141)) H Note that the approximation cov (u, w) ~ 0 is generally not valid and can therefore not be used to simplify the equations 20 RECURSIVE EQUATIONS The estimate can be reformulated into recursive equations, which are easier to implement in a real-time system with limited capacity, Ak = (1 - @) - A ,, _, + gu, f (15a) Bk = (1-.f ) -B ,, _, + ß-w, § (15b) Ek = (1_8) 'Ek_1 + 8'uk'ík Åh P4359SE00p§W fl nl | 60C 2012-04-13vG.9 10 15 20 25 17 Fk = (1-s) -Fk_1 + s-wk-ík (156) where O <s <1 _ Note that E and F must be calculated for the current in each output, but that A, B, D depends only on the zero point voltage which is common to all outputs.
    GENERAL LOW PASS FILTERING It is understood that the recursive formulation can be interpreted as a first order llR filter (llR = infinite impulse response) of the input signal, where the input signals are: Input A = u: InputB = w: InputD = wk uk InputE = ukik InputF = wkík With this approach, it is possible to generalize and use an arbitrary filter with low-pass properties for estimation. Figures 4.1 - 4.3, below illustrate an implementation in the simulation package SlMULlNK ©, where the general approach with low-pass filtering has been used.
    SCALING OF THE DETERMINANT However, a classic problem with the above formulation is the division with the determinant H. If the determinant becomes very small and approaches zero, then the accuracy of the estimation becomes poor. It is important to note that the determinant depends only on the voltage u and its derivatives that are common to all outputs. Therefore, the determinant is identical for all outputs and can be seen as a common scaling factor that is always positive, or zero (positive semidefinite). This follows from which is always positive, or zero, because leuffllêwflzlä fl kwklz- Ah PKJSQSEN p! IN fl nlldoc 2012-04-13 v91. 9 10 15 20 25 30 18 Therefore, the determinant has no effect on the sign of the estimated parameters GAM and Ûm.
    The determinant H goes towards zero, when the zero point voltage decreases towards zero. The determinant can also go to zero if the estimation matrix becomes singular, which means that rows and columns are linear combinations of each other. In the subject area of system identification, a singular matrix means that not all parameters can have an unambiguous value from a given amount of data. In a sense, the determinant reflects the quality or value of the data set to be able to determine the unknown parameters. To avoid numerical problems, the determinant should be limited to a minimum value.
    To determine the faulty output, it is sufficient to know the sign of the estimated parameters G and C. The current of the coil is included as an additional output in the estimation.
    The estimated coil current parameter is used to enable the error output estimation. The criterion for error indication is based on estimated capacitance if Cummmg X <Cco fl 2 Output X is faulty (A) The criterion for error indication based on estimated conductance is if Gumamg X <Gco fl and for all other outputs => Output X is faulty (B) The criteria ( A) and (B) can be combined into logical conditions that can set a memory circuit that remembers the faulty output.
    IMPLEMENTATION The algorithm has been implemented in MATLAB I SlMULlNK and has been tested with disturbance registrations from southern Sweden, recorded from 7 August 2010 to 1 April 2011. Figure 10 shows the three steps in the implementation. In the implementation, the variables A to F are calculated with a general low-pass filter EQ.
    ”= A @> NI P4359SE00 p! sv flnaldøc 2012-04-13 v01 9 10 15 20 25 30 19 where B (q) and A (q) are polynomials in the shift operator q. The filters can usually be constructed with MATLAB © and its toolbox for signal processing.
    The specific choice B (q) = (1 + q "1 + q_2 + ... + q" "+ 1) A (q) = 1 corresponds to the derivation in equations (2) - (9).
    DESCRIPTION OF THE FUNDAMENTAL FREQUENCY METHOD Figure 11 is a flow chart of the fundamental frequency method. For some ground faults, such as intermittent, the mains goes through a sequence of switching where the fault is connected for a short time and then disconnected.
    The underlying problem that the method addresses is that watmetric methods (active power) only provide correct information about error direction, when the error is connected. For a network without any connected faults, the direction of active power cannot be used to identify the faulty output.
    During a sequence of intermittent errors, the error is connected for a short time, where the duration is typically shorter than a fundamental tone cycle, and then the error is disconnected for a significantly longer period of time, and then the sequence is repeated. The dynamics of the zero point voltage are determined by the resonant circuit shown in Figure 2, which, if disturbed from the equilibrium point, will show a slightly attenuated oscillation. So even if the fault is only connected for a short duration, say 5-10 ms, the zero voltage can oscillate more than 100 ms. The results from wattmetric methods, or methods based on active power, are only reliable when the fault is connected. When the fault is disconnected, the voltage source in the equivalent circuit in Figure 2 is also disconnected, and there is only a free oscillation between the Petersen coil and the mains outputs, all of which are healthy.
    The basic idea is to use conventional calculation of phase vectors at fundamental frequency. The amount of the zero point voltage is used to determine whether the fault is connected or not. A decreasing amount of this phase vector is used as a decision criterion to determine whether the fault is disconnected or connected. A decreasing amount is identified by Ah NSSQSEOO po sv Ilnlldoc 2012-04-13 m. B 10 15 20 25 30 20 check if the derivative of the amount of the voltage is negative. If the fault is not connected, wattmetric methods give unreliable results, therefore appropriate measures should be taken in the relay protections, e.g. disable (block) algorithms that use wattmetric information. For example, a blocking signal can be sent to timers and trip counters.
    By including this modification, it is possible to determine when the result of wattmetric methods is reliable. The modification can be used to improve existing wattmetric relay protection for a simple output, for example by stopping counting of the trip counter when the method says that the fault is disconnected. The method can also be used to make a coordinated, coordinated, relay protection for all outputs in the mains station. Consider a common platform for relay protection, which measures all sum currents in a substation. When the method classifies the fault as switched on, the output with the most negative active effect is determined as a faulty output, which can be set in a memory circuit. The memory circuit retains the information about the faulty output until a new fault occurs.
    The principle is illustrated in Figure 13D.
    Note that the direction of active power depends on the definition of what is positive power direction; some electric power companies believe that the positive power direction is in the direction of the busbar, while other electric companies have the opposite. The important knowledge is that all healthy outputs have the same power direction and that the faulty output has the opposite power direction.
    The next section presents two application examples, which illustrate how the method is used. The first example is a single transient earth fault, where the transient method works really well. The second example is chosen to illustrate that sometimes the transient method receives insufficient information and must be supplemented with the modified wattmetric method.
    EXAMPLE 1 - A SINGLE TRANSIENT EARTH ERROR This example illustrates the function of the transient method which estimates the conductance G and the capacitance C. The disturbance has been recorded in a network station in southern Sweden and has a sampling speed of 50 kHz. Figure Ah PßSQSEOO p! lv fi nll flfl c 2012-04-13 var. 9 10 15 20 25 30 21 12A shows an overview and the three graphs are (top to bottom): phase voltage, zero point voltage, and sum current, respectively. The lower graph shows the faulty output, along with two healthy outputs and the current through the zero point equipment.
    The zero voltage shows an initial transient and then decreases, the currents of the outputs show a large initial transient and then decrease to small amounts. This indicates that the earth fault has only been connected for a short time, such as a couple of milliseconds.
    Figure 12B is a close-up of the zero current and the output current. Figure 12B shows initial transients, and clearly shows that during the first 1-2 milliseconds, the currents from the outputs are much larger than the coil current (designation "Npkt" in Figure 12B). Figure 12B also illustrates that the coil current has a much slower dynamic response than the output currents.
    The zero voltage, together with the currents for the outputs and the coil, have been used as input signals to the transient method, which has been used to estimate the conductance G and the capacitance C. Figure 12C shows the resulting estimation of G and C. The upper graph shows the estimated conductance G, and during the first 1-2 ms, the faulty output has the largest amount and negative sign. The negative sign can be explained by Figure 2, which tells us that the current to the faulty output has the opposite sign compared to the healthy outputs. The lower graph in Figure 12C shows the estimated capacitance C. During the time period 341 -350 ms, the estimated capacitance for all healthy outputs is positive and corresponds well to the expected values for these outputs.
    The lower graph in Figure 12C is used to compare the C of the coil with the faulty output. During the first milliseconds, the estimated capacitance, C, for the coil current slowly increases, and is very small compared to the estimated capacitances, C, for the healthy outputs. This explains why the faulty output has a high negative value on the capacitance during the first milliseconds. The long- Oh PQSQSEOO pe lv Ilnaldoc 2012-04-13 vor. 9 10 15 20 25 30 22 the same dynamics in the coil current are reflected in the slow dynamics of the estimated capacitance, C, of the coil.
    The idea is to use the estimated capacitance C of the coil, to extract reliable information from the transient. The rule becomes simple, an output is considered faulty if its estimated capacitance is more negative than the estimated capacitance of the current through the zero point equipment. Another criterion is that the capacitance must also be less than a threshold value, which is there to handle the situation when all estimated capacitances are very close to zero.
    EXAMPLE 2 - INTERMITTENT SOIL ERROR This example illustrates a situation where the modified phase vector method gives more reliable results than the transient method. The disturbance has been recorded in a network station in southern Sweden and the sampling speed was 50 kHz.
    Figure 13A shows a close-up and the two graphs are (top to bottom), zero point voltage, and sum current. The lower graph shows the current in the faulty output, along with two of the healthy outputs and the zero current.
    The characteristic feature of this disorder is five strokes in a time period of less than one second. The zero point voltage rises rapidly, and then decreases, but before it has reached zero, a new blow occurs which increases the zero point voltage. You can also see that the current through the coil is dominated by a 50 Hz component, which does not disappear between the strokes.
    In this situation, it is possible that the decision criterion for the transient method is not met during the later beats, and therefore no information is added from the second and subsequent beats. Note that this does not mean that the transient method provides incorrect information, it simply says that the transient information is insufficient to make a safe decision.
    The reason why the transient method only gives reliable results for the first beat, and not for subsequent beats, is that transients in the output currents have insufficient amplitude to exceed the 50 Hz component of the zero point current created by previous beats. Oh PQSQSEOO p! sv flnaldoc 2912-04-13 vil, 9 10 15 20 25 30 23 Figure 13B shows a close-up of the second type for the time period, 540 ms to 620 ms. It clearly shows that the 50 Hz component of the coil current always has a larger amplitude than transients in the currents of the outputs, in other words, the sum currents of the outputs are always smaller than the coil current.
    This is also reflected in the estimation of the transient values for G and C. Figure 13C shows that the coil current receives the largest amounts for the estimation of both G and C. During this time period, it is therefore uncertain whether the estimated capacitance, C, for the faulty output, will be more negative than the estimated capacitance C for the coil current.
    So in the described situation, there is a clear presence of a 50-Hz component that dominates, and less transient information. In this case, the modified wattmetric method is a useful backup solution. The upper graph in Figure 13D shows the amount of the zero vector phase vector, which is calculated by a conventional DFT method. The middle graph in Figure 13D shows the derivative of the amount of the zero point voltage. In the graph there is also an exemplary threshold value, which is used to find the negative voltage derivative.
    The threshold value should be slightly less than zero, so that permanent errors are not blocked. But at the same time, the threshold value must be selected, so it detects and blocks when the amount of the zero point voltage decreases with a slow time constant.
    The lower graph in Figure 13D shows the active power (wattmetric) for the healthy and faulty output. It clearly shows that as long as the derivative of the amount of the voltage exceeds the threshold value, the active effect for the faulty output is negative and has the largest amount. By combining the criterion for active power, with the derivative of the zero point voltage amount, a reliable and reliable way of determining the faulty output is obtained. Although certain illustrative embodiments of the invention have been specifically described, it is to be understood that various other modifications may be readily apparent to one skilled in the art without departing from the spirit of the invention. Accordingly, it is not intended that the scope of the appended claims be limited by this specification, but that the claims be construed as also covering ÅH PÅSSOSEOO at SV HHILGOC 2Û | All the equivalent embodiments of the invention which will be apparent to those skilled in the art to which the invention pertains. Oh PGGQSEOO p! lv fl nil fi øc ZOIZ-Oß-IS vlrr 9
    权利要求:
    Claims (7)
    [1] 1. A method for detecting earth faults in three-phase electric powerdistribution networks, including continuously measuring at least currents infeeders of all phases and a voltage appearing at a neutral point, andinitiating fault detection if any measured quantities exceed a predefinedthreshold value, c h a ra c t e ris e d by the steps: if transient information exceeds low frequency information, use the steps: a) estimating capacitance and conductance of each feeder, b) estimating capacitance and conductance of the neutralpoint, c) determining a feeder as faulty, either when thecapacitance of the feeder is lower than the capacitance ofthe neutral point device, or when the conductance of thefeeder is lower than the conductance of the neutral pointdevice and lower than the conductance of all otherfeeders; if transient information does not exceed low frequency information, use thesteps: d) determining active power of each feeder and e) determining a feeder as faulty when the active power of said feeder exceeds a predefined value.
    [2] 2. A method as claimed in claim 1, further including the steps:if transient information does not exceed low frequency information,f) determining the magnitude of the neutral point voltagephasor,g) determining a derivative of said magnitude, andh) using active power to determine a fault direction of eachfeeder, if said magnitude exceeds a preset threshold and if said derivative is positive. Åh P4359SE00 finaLdoc 2011-06-14 ver. 5 26
    [3] 3. A method as claimed in anyone and any combination of the precedingclaims including the steps: setting a memory element of each feeder with the determined fault direction,using a content of said memory element if the magnitude of the neutral pointvoltage phasor is over a preset threshold value and the derivative of the magnitude of said phasor is below a negative threshold
    [4] 4. A method as claimed in anyone and any combination of the precedingclaims and performing step d) by: calculating fundamental frequency phasors of a residual current formed bythe sum of phase currents for each feeder and by calculating fundamental frequency phasor of the neutral point voltage.
    [5] 5. A method as claimed in anyone and any combination of the precedingclaims where the value of conductance G and capacitance C for each feederand a neutral point device are estimated by the steps: a) Sample the neutral point voltage, notation uk andresidual current for feeder m, notation imsk , where k is thesample index; b) Calculate the voltage derivative approximation, notation wk from wk w âwk -uk_1) where h is the sampling interval;c) Use the latest n+1 samples (one extra sample is needed for wk) to calculate the following intermediate variables n 2 n 2 nAIZI/lk Bïzwk DIZI/lkwkk=1 k=1 k=1 n nEm I Zukímk Fm I ZWkl-mk k=1 k=1 d) Calculate the intermediate variable Hfrom H=AB-D2; Åh P4359SE00 finaLdoc 2011-06-14 ver. 5 27 e) Calculate the estimated value of conductance G and capacitance C, for feeder with index m, or neutral pointdevice, from Gm ï BEm -DFm öm ï AFm -DEm ;H H
    [6] 6. A method as claimed in anyone and any combination of the preceding claims where the value of conductance G and capacitance C for each feeder and a neutral point device are estimated by the steps: Åh P4359SE00 finaLdoc 2011-06-14 ver. 5 a) b) d) e) Sample the neutral point voltage, notation uk andresidual current for feeder m, notation imsk , where k is thesample index; Calculate the voltage derivative approximation, notationwk from wk w âwk -uk_1) where h is the samplinginterval; Calculate the following intermediate variables 2 xEm I UkímJf XF", I WkímJfÃUse any kind of discrete low pass filter, that can be characterized by a rational transfer function, such as, HLPW) =s, where q is the shift operator, andALP q) BLP(q) and ALP(q) are polynomials in q.Calculate the filtered intermediate variablesA=Hwo)m B I HLP(q)'xB D I HLP(q)'xD Em I HLP(q)'xE,,, Fm :HLP(q)'xF,,,§ 28 f) Calculate the filtered intermediate variable HfromH=AB-D2;g) Calculate the estimated value of conductance G andcapacitance C, for feeder with index m, or neutral point5 device, from Gm : BEm -DFm öm ï AFm -DEm ;H H
    [7] 7. A method as claimed in claim 6, where all steps are taken in continuoustime. Åh P4359SE00 finaLdoc 2011-06-14 ver. 5
    类似技术:
    公开号 | 公开日 | 专利标题
    SE1150535A1|2012-12-15|A method for detecting earth faults
    Zhalefar et al.2015|A high-speed adaptive single-phase reclosing technique based on local voltage phasors
    EP2485354B1|2020-08-12|Protection System for an Electrical Power Network Based on the inductance of a network section
    Brahma2007|Distance relay with out-of-step blocking function using wavelet transform
    Farughian et al.2018|Review of methodologies for earth fault indication and location in compensated and unearthed MV distribution networks
    Hänninen2001|Single phase earth faults in high impedance grounded networks: characteristics, indication and location
    RU2631025C2|2017-09-15|Detection of direction of weakly resistant short circuit to earth of average voltage with help of linear correlation
    US10976357B2|2021-04-13|Method and device for determining a fault location of a ground fault relating to a line of a three-phase electrical energy supply network having a neutral point without low-resistance grounding
    CN103840437B|2016-06-15|The quick diagnosis of power distribution network ferromagnetic resonance and singlephase earth fault and processing method
    RU2633433C2|2017-10-12|Directional detection of fault in network, in particular, in system with grounded compensated or insulated neutral
    US20110074436A1|2011-03-31|Identification of false positives in high impedance fault detection
    CN102520318A|2012-06-27|Fault recognition method for electric transmission line
    Han et al.2014|Fault location on a mixed overhead and underground transmission feeder using a multiple-zone quadrilateral impedance relay and a double-ended travelling wave fault locator
    Dán et al.2010|Comparison of different methods for earth fault location in compensated networks
    KR20180008987A|2018-01-25|Apparatus and method for discriminating fault in gas insulated switchgear system
    Balcerek et al.2012|Centralized substation level protection for determination of faulty feeder in distribution network
    CN103454561B|2015-11-18|A kind of one-phase earthing failure in electric distribution network localization method
    Rashad et al.2020|Adaptive single-end transient-based scheme for detection and location of open conductor faults in HV transmission lines
    Yi et al.2017|An novel in-service cable dielectric loss measurement
    Chatterjee et al.2020|Dual use line relays to improve power swing deblocking function
    de Miguel et al.2013|Implementation of a digital directional Fault Passage Indicator
    Dán et al.2010|Fault location and compensation of the harmonic content of the residual fault current during single-phase to ground faults in compensated networks
    EP1139539B1|2007-04-25|Method for determining the electrical insulation state of the sending end of an electric network
    Chen et al.2002|Developments in directional power line protection using fault transients
    KR20100008197A|2010-01-25|Method for discriminating fault line and phase in ungrounded distribution system
    同族专利:
    公开号 | 公开日
    EP2686691A1|2014-01-22|
    SE536143C2|2013-05-28|
    EP2686691B1|2018-08-01|
    WO2012171694A1|2012-12-20|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    CN108808850A|2017-11-28|2018-11-13|电子科技大学|Intelligent substation based on IEC61850 alerts intelligent analysis method|IES970641A2|1997-08-28|1999-02-24|Electricity Supply Board|Fault detection apparatus and method of detecting faults in an electrical distribution network|
    FI117258B|1998-11-02|2006-08-15|Abb Oy|Ground-fault protection of the electricity network|
    FI106985B|1999-09-23|2001-05-15|Abb Substation Automation Oy|Method for Determining the Distance to Grounding Errors in Electric Distribution Networks by Ring Coupling-Based Calculation|
    US6573726B1|2000-05-02|2003-06-03|Schweitzer Engineering Laboratories, Inc.|Sensitive ground fault detection system for use in compensated electric power distribution networks|
    EP1598674B1|2004-05-18|2006-08-23|Trench Austria GmbH|Method of displaying a high resistance ground leak in a three phase network|
    CN101858948B|2009-04-10|2015-01-28|阿海珐输配电英国有限公司|Method and system for carrying out transient and intermittent earth fault detection and direction determination in three-phase medium-voltage distribution system|CN103454555B|2013-08-02|2016-03-02|西安交通大学|Based on the one-phase earthing failure in electric distribution network Section Location of parameter identification|
    CN103529316B|2013-08-15|2016-09-21|国家电网公司|A kind of method for comprehensive detection of power system high resistance earthing fault|
    CN103490405B|2013-09-29|2015-09-23|张家港智电柔性输配电技术研究所有限公司|A kind of autocontrol method of adjustable capacitive-based arc-suppression coil|
    EP2866039B1|2013-10-25|2016-05-25|ABB Technology AG|Multiphase switch system with a short circuit connection|
    CN103675536B|2013-12-03|2016-09-21|昆明理工大学|A kind of thunderbolt interference identification method utilizing instantaneous power and fitting a straight line|
    CN103675605B|2013-12-11|2016-03-09|湖南大学|A kind of power distribution network earth fault line selection method based on the correlation analysis of fault-signal transient state|
    CN103777103B|2014-02-18|2016-03-30|南京卓实电气有限责任公司|A kind of arc suppression coil state online test method and device|
    CN104753041B|2015-03-20|2017-11-24|北京四方继保自动化股份有限公司|A kind of phase to phase impedance synchronizing calculation method for considering arc light Resistance Influence|
    FI126434B|2015-06-03|2016-11-30|Jyväskylän Energia Oy|Procedure for earth-fault protection in a three-phase mains|
    CN104991144B|2015-07-17|2017-12-08|国网山东省电力公司电力科学研究院|Generator arc suppression coil compensation recognition methods based on transfer overvoltage characteristic frequency|
    SE539916C2|2016-05-11|2018-01-16|Dlaboratory Sweden Ab|Method and device for disconnecting faults in mains|
    WO2017216673A1|2016-06-13|2017-12-21|Electrical Grid Monitoring Ltd.|A method and system for dynamic fault detection in an electric grid|
    US10223906B2|2017-01-23|2019-03-05|Florida Power & Light Company|Open neutral detection|
    CN107831378B|2017-09-13|2020-06-30|梅素真|Device and method for detecting compensation effect of arc suppression coil|
    EP3499252B1|2017-10-18|2020-09-09|China Electric Power Research Institute Company Limited|Single-phase-to-ground fault detection method and device based on monitoring of changes of electric field intensities|
    CN108061842B|2017-12-14|2020-04-10|华北电力大学|Single-phase earth fault positioning method and system for small current grounding system|
    CN108037417B|2017-12-18|2020-01-03|国网宁夏电力有限公司电力科学研究院|Quick detection method for faults of ungrounded system|
    CN111103500A|2018-10-09|2020-05-05|长沙理工大学|Resonance grounding power distribution network ground parameter measurement method considering damping resistance|
    CN109298231B|2018-10-22|2020-12-01|许继变压器有限公司|Current detection method and device for resonance capacitor of arc suppression coil|
    EP3780304A1|2019-08-12|2021-02-17|ABB Schweiz AG|Handling of earth faults in high impedance grounded power systems|
    法律状态:
    优先权:
    申请号 | 申请日 | 专利标题
    SE1150535A|SE536143C2|2011-06-14|2011-06-14|Method for detecting earth faults in three-phase electric power distribution network|SE1150535A| SE536143C2|2011-06-14|2011-06-14|Method for detecting earth faults in three-phase electric power distribution network|
    EP12723836.8A| EP2686691B1|2011-06-14|2012-04-26|A method for detecting earth faults|
    PCT/EP2012/057616| WO2012171694A1|2011-06-14|2012-04-26|A method for detecting earth faults|
    [返回顶部]