Abnormality detection apparatus and abnormality detection method in power supply system

Abstract

To provide an abnormality detection device and an abnormality detection method that are capable of detecting an abnormality in a power distribution line in a power supply system with a current amount smaller than that of an overcurrent relay.SOLUTION: An abnormality detection device 20 includes a current measuring unit 21 that measures a current flowing through a power distribution line 4, a feature calculation unit 24 that calculates a current feature amount from each of a plurality of frames having different start timings in a time waveform of a numerical value related to the current measured by the current measuring unit 21, and a determination unit 25 that determines an abnormality in the power distribution line 4 on the basis of the current feature amount. The current feature amount is a numerical value that changes according to the time waveform and rises or falls when the time waveform indicates an abnormality in the power distribution line 4. The determination unit 25 determines an abnormality in the power distribution line 4 on the basis of the magnitude relationship between the current feature amount calculated by the feature calculation unit 24 and a threshold value.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present disclosure relates to an abnormality detection device and an abnormality detection method for a power supply system. [Background technology]

[0002] Patent Document 1 discloses a protection device for dealing with a short circuit accident in a power supply system. The protection device includes a harmonic supplying means, a current measuring means, a voltage measuring means, a current component extracting means, a voltage component extracting means, a calculation means, and a determination means. The harmonic supplying means supplies harmonics having a frequency higher than a predetermined frequency to a busbar to which an AC voltage of a predetermined frequency is supplied. The current measuring means measures a current value of a distribution line connected to the busbar. The voltage measuring means measures a voltage value of the busbar or the distribution line. The current component extracting means extracts a current component, which is a Fourier coefficient of a harmonic, from a time-varying waveform of the current value measured by the current measuring means. The voltage component extracting means extracts a voltage component, which is a Fourier coefficient of a harmonic, from a time-varying waveform of the voltage value measured by the voltage measuring means. The calculation means calculates an impedance or an admittance as a calculated value from the current component extracted by the current component extracting means and the voltage component extracted by the voltage component extracting means. The determining means determines whether or not a short circuit has occurred in the distribution line based on the calculated value.

[0003] Patent Document 2 discloses a protection device and a protection method for a power distribution system in which an inverter is used as a main power source. This protection device detects a short circuit accident in a power distribution system in which an inverter as a main power source is connected to a busbar and performs protection. This protection device receives as input a busbar voltage detection signal that detects the busbar voltage and a distribution line current detection signal that detects the currents flowing through a plurality of distribution lines connected to the busbar. This protection device includes a first protection element, a second protection element, and a third protection element. The first protection element operates with a time limit when a logical product is established between the output of a first overcurrent relay that is set to a current value at which the inverter can operate without stopping due to an overcurrent, and the output of an undervoltage relay that detects a drop in the line voltage of a plurality of distribution lines. The second protection element has an instantaneous element of a second overcurrent relay and operates instantaneously when the input current exceeds the set overcurrent level. The third protection element has a time-limit element of the second overcurrent relay, and operates with a time limit when the input current exceeds a set overcurrent level. This protection device determines that a short circuit has occurred when the logical sum condition of the operations of the first to third protection elements is satisfied. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2018-183034 A [Patent Document 2] JP 2021-175246 A Summary of the Invention [Problem to be solved by the invention]

[0005] In recent years, microgrids have become known as a form of power supply system. A microgrid is a small-scale power supply system that integrates power generation, storage, distribution, and consumption within a certain area. A microgrid combines renewable energy sources such as solar or wind power generation, storage devices, and diesel generators to provide a stable power supply to a specific area or facility. Microgrids can also be operated independently from large-scale power grids, and can continue to supply power even if the power grid is shut down due to a disaster or other reason. In addition, microgrids can achieve efficient power use, energy savings, and CO2 reduction by adjusting the power supply and demand balance within a region. 2 It can also contribute to reducing emissions.

[0006] In a power supply system such as the above-mentioned microgrid, power is supplied from a bus connected to a power supply device through a plurality of distribution lines to power loads such as electrical devices connected to each distribution line. Conventionally, in such a power supply system, when an abnormality such as a short circuit occurs in any of the distribution lines, the distribution line is cut off by an overcurrent relay provided in the abnormal distribution line to protect other parts of the power supply system. Since an overcurrent relay generally operates when a current amount several times larger than a normal current flows, a current load several times larger is applied to the power supply system before the overcurrent relay cuts off the distribution line. If the power supply device can supply such a current, there is no problem. However, in the case of an inverter power supply used in many renewable energy power sources, for example, when the allowable current load of the power supply device is relatively small, the abnormality in the distribution line affects other parts of the power supply system before the overcurrent relay cuts off the distribution line.

[0007] The present disclosure has been made in consideration of such problems, and aims to provide an abnormality detection device and an abnormality detection method that are capable of detecting abnormalities in power distribution lines in a power supply system with a smaller amount of current than an overcurrent relay. [Means for solving the problem]

[0008] [1] The anomaly detection device according to the present disclosure is an anomaly detection device provided in a power supply system including a bus connected to a power supply device, a distribution line connected to the bus, and a power load connected to the distribution line, and includes a current measurement unit that measures a current flowing through the distribution line, a feature calculation unit that calculates a current feature amount from each of a plurality of frames having different start timings in a first time waveform that is a time waveform of a numerical value related to the current measured by the current measurement unit, and a determination unit that determines an anomaly in the distribution line based on the current feature amount. The current feature amount is a numerical value that changes according to the first time waveform and rises or falls when the first time waveform indicates an anomaly in the distribution line. The determination unit determines an anomaly in the distribution line based on a magnitude relationship between the current feature amount calculated by the feature calculation unit and a first threshold value.

[0009] The time waveform of the current value forms characteristic waveforms corresponding to various events, such as distribution line abnormalities such as short circuits and ground failures, and normal operations such as turning on and off a power load. The anomaly detection device described in [1] above calculates a current feature value, which is a numerical value that rises or falls when an abnormality occurs in the distribution line, from the first time waveform, and determines an abnormality in the distribution line based on the current feature value. This makes it possible to detect an abnormality in a distribution line in a power supply system with a smaller current amount than an overcurrent relay.

[0010] [2] In the anomaly detection device of [1] above, the numerical value related to the current may be at least one numerical value selected from the group consisting of the current value, frequency, phase, and rate of change of the current value of the current measured by the current measuring unit. Regardless of which numerical value related to the current is, a current feature value that increases or decreases when an anomaly occurs in the power distribution line can be calculated from the first time waveform.

[0011] [3] In the abnormality detection device of [1] or [2] above, the abnormality of the distribution line may be a short circuit or a ground fault of the distribution line. When a short circuit or a ground fault occurs in the distribution line, a characteristic waveform is formed in the time waveform of the numerical value related to the current. Therefore, by calculating the current feature amount from the first time waveform, it is possible to accurately determine a short circuit or a ground fault in the distribution line.

[0012] [4] In the abnormality detection device of [1] to [3] above, the feature calculation unit may acquire numerical values related to the current at each of the N timings included in each frame to generate a sequence of N terms, add or subtract adjacent terms included in the sequence to update the sequence, and thereafter, repeat the addition or subtraction of adjacent terms and use the sum of all terms of the finally generated sequence as the current feature amount. According to the research of the present inventor, the current feature amount calculated in this way significantly increases or decreases when an abnormality occurs in the distribution line as compared with normal operations such as the energization and de-energization of the power load. Therefore, it is possible to accurately determine the abnormality of the distribution line.

[0013] [5] In the abnormality detection device of [1] to [3] above, the feature calculation unit may store in advance an adjustment sequence including N adjustment coefficients. The feature calculation unit may acquire numerical values related to the current at each of the N timings included in each frame to generate a sequence of N terms, multiply each of the N numerical values included in the sequence by each of the N adjustment coefficients, then add or subtract adjacent terms to update the sequence, and thereafter, repeat the multiplication of each of the N adjustment coefficients and the addition or subtraction of adjacent terms, and use the sum of all terms of the finally generated sequence as the current feature amount. According to the research of the present inventor, the current feature amount calculated in this way significantly increases or decreases when an abnormality occurs in the distribution line as compared with normal operations such as the energization and de-energization of the power load. Therefore, it is possible to accurately determine the abnormality of the distribution line.

[0014] [6] The abnormality detection device of [1] to [5] above may further include a wiring breaker that breaks the distribution line when an abnormality of the distribution line is detected by the determination unit. In that case, it is possible to protect other parts of the power supply system except for the distribution line where the abnormality has occurred.

[0015] [7] In the anomaly detection device according to [1] to [6] above, a portion including a start timing of one of the adjacent frames may overlap with a portion including an end timing of the other frame in time. In this case, the feature amount can be calculated from a time waveform over a longer period, and the current feature amount can be calculated at short time intervals. Therefore, an anomaly in the power distribution line can be detected quickly and with high accuracy.

[0016] [8] The anomaly detection device according to [1] to [7] above may further include a voltage measurement unit that measures a voltage in the power distribution line. The feature calculation unit may further calculate a voltage feature amount from each of a plurality of frames having different start timings in a second time waveform that is a time waveform of a numerical value related to the voltage measured by the voltage measurement unit. The voltage feature amount may be a numerical value that changes according to the second time waveform and increases or decreases when the second time waveform indicates an anomaly in the power distribution line. The determination unit may further determine an anomaly in the power distribution line based on a magnitude relationship between the voltage feature amount calculated by the feature calculation unit and a second threshold value. In this case, an anomaly in the power distribution line can be determined with higher accuracy.

[0017] [9] The anomaly detection method according to the present disclosure is an anomaly detection method in a power supply system including a bus connected to a power supply device, a distribution line connected to the bus, and a power load connected to the distribution line, and includes a current measurement step of measuring a current flowing through the distribution line, a feature calculation step of calculating a current feature from each of a plurality of frames having different start timings in a first time waveform, which is a time waveform of a numerical value related to the current measured by a current measurement unit, and a determination step of determining an anomaly in the distribution line based on the current feature. The current feature is a numerical value that changes according to the first time waveform and rises or falls when the first time waveform indicates an anomaly in the distribution line. In the determination step, an anomaly in the distribution line is determined based on the magnitude relationship between the current feature calculated by the feature calculation unit and a first threshold value. According to this anomaly detection method, as with the anomaly detection device of the above [1], it is possible to detect an anomaly in a distribution line in a power supply system with a current amount smaller than that of an overcurrent relay. Effect of the Invention

[0018] According to the present disclosure, it is possible to provide an abnormality detection device and an abnormality detection method that are capable of detecting an abnormality in a power distribution line in a power supply system with a smaller amount of current than an overcurrent relay. [Brief description of the drawings]

[0019] [Figure 1] FIG. 1 is a diagram illustrating a schematic configuration of a power supply system. [Diagram 2] FIG. 2 is a diagram showing a schematic internal configuration of one customer premises. [Diagram 3] FIG. 3 is a graph showing an example of a time waveform of a current-related value. [Figure 4] FIG. 4 is a diagram showing an image of calculation of feature amounts. [Diagram 5] FIG. 5 is a diagram intuitively illustrating the detection feature vector. [Figure 6] FIG. 6 is a diagram showing an image of calculation of feature amounts. [Figure 7]FIG. 7 is a diagram conceptually showing the operation of the discrimination calculation unit. [Figure 8] FIG. 8 is a flow chart illustrating an example method for building an identification dictionary. [Figure 9] FIG. 9 is a flow chart illustrating an anomaly detection method according to one embodiment. [Figure 10] FIG. 10 is a graph showing an example of the simulation. [Figure 11] FIG. 11 is a graph showing another example of the simulation. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0020] Hereinafter, an embodiment of an anomaly detection device and an anomaly detection method according to the present disclosure will be described in detail with reference to the accompanying drawings. In the description of the drawings, the same elements are given the same reference numerals and duplicated description will be omitted.

[0021] Fig. 1 is a diagram showing a schematic configuration of a power supply system 1. As shown in Fig. 1, the power supply system 1 includes a bus 3 connected to an external power source 2, a plurality of distribution lines 4 connected to the bus 3, and a plurality of power loads 5 connected to each of the plurality of distribution lines 4. The power supply system 1 further includes one or more power feed lines 6 connected to the bus 3, and a plurality of power generation devices 7 connected to each of the one or more power feed lines 6. The external power source 2 and the power generation devices 7 are examples of power supply devices in this embodiment. The power supply system 1 configures a microgrid.

[0022] The external power source 2 is a power source of a power company that operates the power supply system 1. The power supplied from the external power source 2 is commercial AC power. A switch 9 is connected between the external power source 2 and the bus line 3 for interruption.

[0023] One or more power generation devices 7 are distributed power sources that generate power independently. Each power generation device 7 includes an independent power source 71, such as a solar cell or a storage battery, and a power conditioning system (PCS) 72. The PCS 72 includes an inverter for converting DC power from the independent power source 71 into commercial AC power. Each power generation device 7 is connected to a corresponding power supply line 6 via a transformer 14. Each power supply line 6 is provided with a switch 13 for cutting off the power supply line 6.

[0024] Each of the multiple distribution lines 4 receives power from an external power source 2 and / or a power generation device 7 via a bus bar 3 and supplies the power to a corresponding power load 5. Each of the multiple distribution lines 4 is provided with a switch 10 (wiring breaker) for breaking the distribution line 4. The switch 10 is, for example, a switch that opens and closes a load current. The switch 10 is, for example, a circuit breaker.

[0025] The multiple power loads 5 are power loads such as electrical equipment provided for each consumer. The power loads include a capacitor, a rectifier, a motor, etc. Each of the multiple power loads 5 is connected to a corresponding distribution line 4 via a transformer 11. Each power load 5 and the corresponding distribution line 4 are provided in the corresponding consumer premises 12.

[0026] When a failure occurs in the external power source 2 due to a disaster or the like, the switch 9 is disconnected and the power supply system 1 operates autonomously. At that time, power from each power generation device 7 is supplied to multiple power loads 5 via the power feeder 6, the bus bar 3, and the distribution line 4 (arrow A in the figure).

[0027] Fig. 2 is a diagram illustrating a schematic internal configuration of one customer premises 12. As illustrated in Fig. 2, the power supply system 1 includes an abnormality detection device 20 in each customer premises 12. The abnormality detection device 20 includes a current measurement unit 21, a voltage measurement unit 22, and an abnormality detection unit 23.

[0028] The current measuring unit 21 measures the current flowing through the power distribution line 4 (one of phase A, phase B, or phase C if the power distribution line 4 is three-phase). A signal indicating the measurement result in the current measuring unit 21 is provided from the current measuring unit 21 to the abnormality detecting unit 23. The voltage measuring unit 22 measures the voltage of the power distribution line 4. A signal indicating the measurement result in the voltage measuring unit 22 is provided from the voltage measuring unit 22 to the abnormality detecting unit 23.

[0029] The abnormality detection unit 23 judges the presence or absence of an abnormality in the power distribution line 4 based on the measurement results in the current measurement unit 21 and the voltage measurement unit 22. The abnormality detection unit 23 has a measurement unit 231, a data pre-processing unit 232, a feature extraction unit 233, an identification calculation unit 234, an identification dictionary 235, a control unit 236, a communication unit 237, and an output unit 238. The data pre-processing unit 232 and the feature extraction unit 233 constitute the feature calculation unit 24. The identification calculation unit 234 and the identification dictionary 235 constitute the determination unit 25.

[0030] The measurement unit 231 is connected to the current measurement unit 21 and the voltage measurement unit 22. The measurement unit 231 receives signals indicating the detection results from the current measurement unit 21 and the voltage measurement unit 22, and samples the signals to convert them into time-series data.

[0031] The data preprocessing unit 232 receives the time series data from the measurement unit 231 and converts the time series data into a format suitable for anomaly detection. Specifically, the data preprocessing unit 232 obtains a time waveform (first time waveform) of a numerical value related to current (hereinafter, referred to as a current-related value) from the time series data. The current-related value is, for example, at least one numerical value selected from a group consisting of a current value, a frequency, a phase, and a rate of change of a current value. The time waveform of the current-related value is obtained as time series data of the current-related value. FIG. 3 is a graph that shows an example of a time waveform of the current-related value. In FIG. 3, the vertical axis represents the amplitude of the numerical value, and the horizontal axis represents time. The data preprocessing unit 232 breaks down the time waveform (time series data) of the current-related value into a plurality of frame data related to a plurality of frames FL having different start timings. At this time, as shown in FIG. 3, a portion including the start timing of one frame FL among the frames FL adjacent to each other in the plurality of frames FL may overlap in time with a portion including the end timing of the other frame FL. The frame data is a sequence of N terms {I 0 ,I 1 ,···,I N} (where N is an integer greater than or equal to 2). n (n=1, . . . , N) is the n-th numerical value counted from the start timing of the frame FL. Hereinafter, this sequence of numbers may be referred to as a "detection feature vector."

[0032] The feature extraction unit 233 extracts a feature (current feature) from each of the frame data of the multiple frames FL. The feature is a numerical value that changes according to the time waveform of the current-related value and increases or decreases when the time waveform indicates an abnormality in the power distribution line 4. An abnormality in the power distribution line 4 is, for example, a short circuit or a fall to the ground in the power distribution line 4. Two examples of the feature will be described below. [First example of features]

[0033] In a first example, the feature extraction unit 233 updates the detected feature vector by adding or subtracting adjacent terms included in the detected feature vector, and thereafter repeats the addition or subtraction of adjacent terms. A specific example is shown below, but the calculation method is not limited to the following example.

[0034] First, the initial detection feature vector F 0 ={I 0 ,I 1 ,···,I N}, the following calculation is performed to obtain the detected feature vector F 1 Request. F 1 ={(I 0 +I 1 ),(I 2 +I 3 ),···,(I 0 -I 1 ),(I 2 -I 3 ),···} Next, the detection feature vector F 1 The following calculation is performed to obtain the detected feature vector F 2 Request. F 2 ={(I 0 +I 1 )+(I 2 +I 3 ),(I 0 -I 1 )+(I 2 -I 3 ),···,(I 0 +I 1 )-(I 2 +I 3 ),(I 0 -I 1 )-(I 2 -I 3 ),···} That is, the detected feature vector after m-th calculation (m is an integer between 1 and M, and M is the number of repetitions) is expressed as F m ={I 0 m ,I 1 m ,···,I N m}, in the (m+1)th calculation, the detected feature vector Fm+1 of F m+1 ={(I 0 m +I 1 m ),(I 2 m +I 3 m ),···,(I 0 m -I 1 m ),(I 2 m -I 3 m ),...}. For example, F 0 = {1,2,2,1}, then F 1 ={3,3,-1,1}, F 2 ={6,0,0,-2}.

[0035] Next, the feature extraction unit 233 extracts the last generated detection feature vector F M The sum of all the terms in the above is used as the feature value. That is, the feature value Ft is calculated by the following formula: Ft=I 0 M +I 1 M +···+I N M For example, if the last generated detection feature vector is F 2 If Ft = {6,0,0,-2}, then Ft = 6 + 0 + 0 - 2 = 4. [Second example of features]

[0036] In the first example, the feature extraction unit 233 calculates the N-term adjustment coefficient Zs 0 ~Zs N The sequence of adjustments Zs={Zs 0 ,Zs 1 ,···,Zs N} is stored in advance. The feature extraction unit 233 then stores the numerical value I 0 ,I 1 ,···,I N Each Nth term adjustment coefficient Zs 0 ,Zs 1, ···, Zs N After multiplying each of them, adjacent terms are added or subtracted from each other to update the detected feature vector. Thereafter, the feature extraction unit 233 uses N adjustment coefficients Zs 0 , Zs 1 , ···, Zs N The multiplication of each and the addition or subtraction of adjacent terms are repeated. A specific example is shown below, but the calculation method is not limited to the following example.

[0037] First, for the first detected feature vector F 0 ={I 0 , I 1 , ···, I N}, the following operations are performed to obtain the detected feature vector F 1 . F 1 ={(I 0 + I 1 )·Zs 0 , (I 2 + I 3 )·Zs 1 , ···, (I 0 - I 1 )·Zs N / 2 , (I 2 - I 3 )·Zs (N / 2)+1 , ···} Next, for the detected feature vector F 1 , the following operations are performed to obtain the detected feature vector F 2 . F 2 ={((I 0 + I 1 ) + (I 2 + I 3 ))·Zs 0 , ((I 0 - I 1 ) + (I 2 - I 3 ))·Zs 1 , ···, ((I 0 + I 1 ) - (I 2 + I 3 ))·Zs N / 2 , ((I 0 - I 1 ) - (I 2 - I 3 ))·Zs(N / 2)+1 ,···} That is, the detected feature vector after m operations is F m ={I 0 m ,I 1 m ,···,I N m}, in the (m+1)th calculation, the detected feature vector F m+1 of F m+1 ={(I 0 m +I 1 m )·Zs 0 ,(I 2 m +I 3 m )·Zs 1 ,···,(I 0 m -I 1 m )·Zs N / 2 ,(I 2 m -I 3 m )·Zs (N / 2)+1 , ···}. Note that in this second example as well, the number of iterations M is arbitrary.

[0038] In the above calculation, the addition term (I 0 m +I 1 m The addition term effectively distinguishes between signals close to zero and signals close to amplitude peaks. The subtraction term (I 0 m -I 1 m The subtraction term effectively identifies small oscillations in a signal, such as those shown by plot P in Figure 4.

[0039] Next, the feature extraction unit 233 extracts the last generated detection feature vector F MThe sum of all the terms in the above is used as the feature value. That is, the feature value Ft is calculated by the following formula: Ft=I 0 M +I 1 M +···+I N M

[0040] Figure 5 shows the detection feature vector F NI This is a diagram intuitively showing the detected feature vector F NI is the feature vector F 0 ~F M In Fig. 5, the first axis represents time t, the second axis represents the number of iterations M, and the third axis represents the detection feature vector F NI FIG. 6 is a diagram showing an image of calculation of the feature amount Ft. The first to third axes in FIG. 6 are the same as those in FIG. 5. The detected feature vector Ft after a predetermined number of repetitions M is M The sum of all the terms (shown in area C in FIG. 6) is the feature quantity Ft.

[0041] Referring again to FIG. 2, the discrimination calculation unit 234 judges an abnormality in the power distribution line 4 based on the feature amount Ft. The discrimination calculation unit 234 judges an abnormality in the power distribution line 4 based on the magnitude relationship between the feature amount Ft calculated by the feature extraction unit 233 and a judgment threshold (first threshold). The judgment threshold is set by using an adjustment coefficient Zs 0 ~Zs N , has a number of different values ​​according to the number of repetitions M and the number of items N, and is stored in the identification dictionary 235.

[0042] 7 is a diagram conceptually showing the operation of the discrimination calculation unit 234. As shown in FIG. 7, the discrimination calculation unit 234 first calculates an adjustment coefficient Zs 0 ~Zs N , the number of repetitions M, the number of terms N, and the feature value Ft are input (block B1 in the figure). Next, the adjustment coefficient Zs 0 ~Zs N Based on the number of repetitions M and the number of terms N, the discrimination dictionary 235 is searched (block B2 in the figure). For example, the adjustment coefficient Zs 0 ~Zs NWhen the number of repetitions M and the number of terms N are certain values, the discrimination calculation unit 234 determines that a short circuit has occurred when the feature value Ft exceeds 15, and determines that the power load 5 should be powered on when the feature value Ft is greater than 10 and less than 15 (block B3 in the figure). 0 ~Zs N When at least one of the number of repetitions M and the number of terms N is different from the above, the discrimination calculation unit 234 determines that a short circuit has occurred if the feature value Ft exceeds 50, and determines that the power load 5 should be powered on if the feature value Ft is greater than 40 and less than 50 (block B4 in the figure). 0 ~Zs N When at least one of the number of repetitions M and the number of terms N is a different value, the identification calculation unit 234 determines that a short circuit has occurred if the feature value Ft exceeds 100, and determines that the power load 5 should be powered on if the feature value Ft is greater than 70 and less than 100 (block B5 in the figure).

[0043] The threshold values ​​stored in the identification dictionary 235 may be determined based on values ​​previously obtained by experiment or simulation as to what values ​​the feature quantity Ft will take when an abnormality such as power-on or short circuit of the power load 5 occurs. FIG. 8 is a flowchart showing an example of a method for constructing the identification dictionary 235. In this example, a case is shown in which the abnormality in the power distribution line 4 is a short circuit in the power distribution line 4. First, frame data including a time waveform (time series data) at the time of the short circuit and frame data including a time waveform (time series data) at the time of power-on of the power load 5 are input (step ST1). Next, the adjustment coefficient Zs 0 ~Zs N Using the feature quantity Ft at the time of a short circuit of the distribution line 4 and the feature quantity Ft at the time of power-on of the power load 5, the number of repetitions M and the number of terms N are calculated (step ST2). Then, it is determined whether the feature quantity Ft at the time of a short circuit is larger than the feature quantity Ft at the time of power-on (step ST3). If the feature quantity Ft at the time of a short circuit is not larger than the feature quantity Ft at the time of power-on (step ST3: NO), the adjustment coefficient Zs 0 ~Zs N, the number of repetitions M, and the number of terms N are changed (step ST4). After that, steps ST2 and ST3 are performed again. If the feature amount Ft at the time of short circuit is greater than the feature amount Ft at the time of power-on (step ST3: YES), the average value of the feature amount Ft at the time of short circuit and the feature amount Ft at the time of power-on is used as a threshold value for classification, and an adjustment coefficient Zs corresponding to the threshold value is calculated. 0 ~Zs N , the number of repetitions M, and the number of terms N are recorded (step ST5). By repeating the above steps ST1 to ST5, the identification dictionary 235 is constructed.

[0044] 2 again. The control unit 236 controls the measurement unit 231, the data preprocessing unit 232, the feature extraction unit 233, the identification calculation unit 234, the identification dictionary 235, the communication unit 237, and the output unit 238. In particular, when the identification calculation unit 234 detects an abnormality in the power distribution line 4, the control unit 236 controls the switch 10 to a non-connected state so as to cut off the power distribution line 4 and separate it from the bus bar 3. In addition, when the identification calculation unit 234 determines that an abnormality such as a short circuit has occurred, the communication unit 237 and the output unit 238 notify the outside of the power supply system 1 of the abnormality such as a short circuit.

[0045] Next, an abnormality detection method according to the present embodiment will be described. FIG. 9 is a flowchart showing the abnormality detection method according to the present embodiment. This abnormality detection method can be implemented, for example, by using the abnormality detection device 20 described above. In this abnormality detection method, first, a current flowing through the power distribution line 4 is measured (current measurement step ST11). Next, in a time waveform (time series data) of a value related to the current (current-related value) measured in the current measurement step ST11, a feature value Ft is calculated from each of a plurality of frames FL having different start timings (feature calculation step ST12). The current-related value is, for example, at least one value selected from a group consisting of a current value, a frequency, a phase, and a rate of change of the current value of the current measured in the current measurement step ST11. The feature value Ft is a value that changes according to the time waveform of the current-related value and rises or falls when the time waveform indicates an abnormality in the power distribution line 4. Next, an abnormality in the power distribution line 4 is determined based on the feature value Ft (determination step ST13). The abnormality in the power distribution line 4 is, for example, a short circuit or a ground fault in the power distribution line 4. In the determination step ST13, an abnormality in the power distribution line 4 is determined based on the magnitude relationship between the feature amount Ft calculated in the feature calculation step ST12 and a threshold value stored in the identification dictionary 235. If it is determined in the determination step ST13 that no abnormality has occurred in the power distribution line 4 (determination step ST13: NO), the above-mentioned current measurement step ST11, feature calculation step ST12, and determination step ST13 are repeated. If it is determined in the determination step ST13 that an abnormality has occurred in the power distribution line 4 (determination step ST13: YES), the switch 10 cuts off the power distribution line 4 in which the abnormality has occurred and separates it from the bus bar 3 (cutting step ST14).

[0046] The effects of the anomaly detection device 20 and the anomaly detection method according to the present embodiment described above will be described. The time waveform of the current-related value forms characteristic waveforms corresponding to various events, such as anomalies in the power distribution line 4, such as a short circuit and a ground fault, and normal operations, such as turning on and off the power load 5. The anomaly detection device 20 and the anomaly detection method according to the present embodiment calculate a feature value Ft, which is a numerical value that rises or falls when an anomaly occurs in the power distribution line 4, from the time waveform (time-series data), and judges an anomaly in the power distribution line 4 based on the feature value Ft. This makes it possible to detect an anomaly in the power distribution line 4 in the power supply system 1 with a current amount smaller than the operating current amount of the overcurrent relay.

[0047] FIG. 10 is a graph showing an example of a simulation. FIG. 10(a) shows time series data of the current flowing through the power distribution line 4. FIG. 10(b) shows the feature value Ft calculated from the time series data of FIG. 10(a) by the first example. FIG. 10(c) shows the feature value Ft calculated from the time series data of FIG. 10(a) by the second example. In FIG. 10(a) to (c), the horizontal axis represents the sample number (sampling frequency 10 kHz). The vertical axis of FIG. 10(a) represents the amplitude of the instantaneous value of the current. The vertical axes of FIG. 10(b) and (c) represent the feature value Ft. FIG. 10 shows the power-on timing T1 of the power load 5 and the short-circuit timing T2 of the power distribution line 4.

[0048] Referring to FIG. 10(a), there is no significant difference between the amount of change ΔJ1 in the current amplitude caused by the power-on of the power load 5 and the amount of change ΔJ2 in the current amplitude caused by the short circuit of the power distribution line 4. Therefore, it is difficult to distinguish between the power-on of the power load 5 and the short circuit of the power distribution line 4 only from the amount of change in the current amplitude. In contrast, as shown in FIG. 10(b) and (c), the feature amount Ft when the power distribution line 4 is short-circuited is much larger than the feature amount Ft when the power load 5 is powered on. Therefore, by setting a threshold value TH having a value between the feature amount Ft when the power load 5 is powered on and the feature amount Ft when the power distribution line 4 is short-circuited, it is possible to accurately distinguish between the power-on of the power load 5 and the short circuit of the power distribution line 4. In addition to the power-on of the power load 5, there are fluctuations in the current value during normal operation, such as the release of the power load 5, which are similar to the power-on of the power load 5. In addition to the short circuit of the power distribution line 4, there are abnormalities, such as the power distribution line 4 falling to the ground, which are similar to the short circuit of the power distribution line 4.

[0049] FIG. 11 is a graph showing another example of the simulation. FIG. 11(a) shows time series data of the current flowing through the power distribution line 4. FIG. 11(b) shows a feature value Ft calculated from the time series data of FIG. 11(a). Note that in (a) to (c) of FIG. 11, the horizontal axis also represents the sample number (sampling frequency 10 kHz). The vertical axis of FIG. 11(a) represents the amplitude of the instantaneous value of the current. The vertical axis of FIG. 11(b) represents the feature value Ft. FIG. 11 shows the power-on timing T1 of the power load 5, the short-circuit timing T2 of the power distribution line 4, and the timing T3 of the intentional current limiting.

[0050] 11(a), the amount of change in current amplitude due to a short circuit in the distribution line 4 is extremely small compared to the amount of change in current amplitude due to power-on of the power load 5 and the amount of change in current amplitude due to current limiting. Therefore, it is impossible to distinguish between power-on and current limiting of the power load 5 and a short circuit in the distribution line 4 from only the amount of change in current amplitude. In contrast, as shown in FIG. 11(b), the feature amount Ft when the distribution line 4 is short-circuited is significantly larger compared to each feature amount Ft when the power load 5 is powered on and current limited. Therefore, by setting a threshold value TH, it is possible to accurately distinguish between power-on and current limiting of the power load 5 and a short circuit in the distribution line 4.

[0051] As described above, the current-related value may be at least one numerical value selected from the group consisting of the current value, frequency, phase, and rate of change of the current value measured by the current measuring unit 21. Regardless of which of these numerical values ​​the current-related value is, the feature value Ft that increases or decreases when an abnormality occurs in the power distribution line 4 can be calculated from the time waveform (time-series data) of the current-related value.

[0052] As described above, the abnormality in the power distribution line 4 may be a short circuit or a fall to the ground in the power distribution line 4. When a short circuit or a fall to the ground occurs in the power distribution line 4, a characteristic waveform is formed in the time waveform of the current-related value. Therefore, by calculating the characteristic amount Ft from the time waveform (time-series data) of the current-related value, it is possible to accurately determine whether or not a short circuit or a fall to the ground has occurred in the power distribution line 4.

[0053] As described above, the feature calculation unit 24 obtains current-related values ​​at each of N timings included in each frame FL and calculates a sequence of N terms (detection feature vector F 0 ), and then the sequence is updated by adding or subtracting adjacent terms contained in the sequence. After that, the addition or subtraction of adjacent terms is repeated to generate the final sequence (detection feature vector F M ) the sum of all terms (I 0 M +I 1 M +···+I N M) may be used as the feature value Ft. According to the inventor's research, the feature value Ft calculated in this way increases or decreases significantly when an abnormality occurs in the power distribution line 4, compared with during normal operations such as turning on and off the power load 5. Therefore, an abnormality in the power distribution line 4 can be determined with high accuracy.

[0054] As described above, the feature calculation unit 24 calculates the N-term adjustment coefficient Zs 0 ~Zs N The sequence of adjustments Zs={Zs 0 ,Zs 1 ,···,Zs N The feature calculation unit 24 may store in advance the N-term sequence (detection feature vector F 0 ) and generate the numerical value I 0 ,I 1 ,···,I N Each Nth term adjustment coefficient Zs 0 ,Zs 1 ,···,Zs N After multiplying each, add or subtract adjacent terms to update the sequence, and then use the Nth term adjustment coefficient Zs 0 ,Zs 1 ,···,Zs N By repeating each multiplication and adding or subtracting adjacent terms, the final generated sequence (detection feature vector F M ) the sum of all terms (I 0 M +I 1 M +···+I N M ) may be used as the feature value Ft. According to the inventor's research, the feature value Ft calculated in this way increases or decreases significantly when an abnormality occurs in the power distribution line 4, compared with during normal operations such as turning on and off the power load 5. Therefore, an abnormality in the power distribution line 4 can be determined with high accuracy.

[0055] As in the present embodiment, the abnormality detection device 20 may include a switch 10 that cuts off the power distribution line 4 when the determination unit 25 detects an abnormality in the power distribution line 4. In this case, it is possible to protect other parts of the power supply system 1, except for the power distribution line 4 in which an abnormality has occurred.

[0056] As in the present embodiment, a portion including the start timing of one frame FL and a portion including the end timing of the other frame FL among adjacent frames FL may overlap in time. In this case, the feature amount Ft can be calculated from a time waveform over a longer period, and the feature amount Ft can be calculated at short time intervals. Therefore, an abnormality in the power distribution line 4 can be detected quickly and with high accuracy.

[0057] [Variations] Next, a modified example of the above embodiment will be described. In the above embodiment, the abnormality of the power distribution line 4 is determined based only on the feature value Ft calculated from the numerical value (current-related value) related to the current measured by the current measuring unit 21. However, the feature value Ft may be further calculated from the numerical value (voltage-related value) related to the voltage measured by the voltage measuring unit 22, and the abnormality of the power distribution line 4 may be further determined based on the feature value Ft. In this case, the data preprocessing unit 232 obtains a time waveform (second time waveform) of the voltage-related value from the time-series data received from the measurement unit 231. The voltage-related value is, for example, at least one numerical value selected from the group consisting of a voltage value, a frequency, a phase, and a rate of change of the voltage value. The time waveform of the voltage-related value is obtained as time-series data of the voltage-related value. As in the case of the current-related value (see FIG. 3), the data preprocessing unit 232 breaks down the time waveform (time-series data) of the voltage-related value into a plurality of frame data related to a plurality of frames FL having different start timings. In this case, a portion including the start timing of one of the adjacent frames FL may overlap in time with a portion including the end timing of the other frame FL. The frame data is created as an N-term sequence (detection feature vector) indicating voltage-related values ​​at each of N timings included in each frame FL.

[0058] The feature extraction unit 233 extracts a feature (voltage feature) Ft from each of the frame data of the multiple frames FL. The feature Ft is a value that changes according to the time waveform of the voltage-related value and increases or decreases when the time waveform indicates an abnormality in the power distribution line 4. An abnormality in the power distribution line 4 is, for example, a short circuit or a ground fault in the power distribution line 4. A method for calculating the feature Ft from the detection feature vector of the voltage-related value is similar to the method for calculating the feature Ft from the detection feature vector of the current-related value (the first or second example described above). The discrimination calculation unit 234 determines an abnormality in the power distribution line 4 based on the magnitude relationship between the feature Ft of the current-related value calculated by the feature extraction unit 233 and a judgment threshold (first threshold) and the magnitude relationship between the feature Ft of the voltage-related value and a judgment threshold (second threshold). For example, the discrimination calculation unit 234 determines that an abnormality has occurred in the power distribution line 4 when both the feature Ft of the current-related value and the feature Ft of the voltage-related value exceed the judgment threshold.

[0059] The anomaly detection device and the anomaly detection method according to the present disclosure are not limited to the above-described embodiment and modification, and various other modifications are possible. For example, in the above-described embodiment and modification, the detection feature vector F of the current-related value is 0 Based on the feature value Ft calculated from the current-related value detection feature vector F 0 The feature vector Ft calculated from the voltage-related value detection feature vector F 0 The feature quantity used to determine an abnormality in the power distribution line 4 is not limited to the current-related value and the voltage-related value, and may be calculated from various other parameters (e.g., impedance or admittance calculated from the voltage value and the current value). In addition, the detection feature vector F of the voltage-related value 0 Alternatively, an abnormality in the power distribution line 4 may be determined based only on the feature value Ft calculated from the above.

[0060] In addition, the current-related value and the voltage-related value are not limited to the numerical values ​​(current value or voltage value, frequency, phase, and rate of change) exemplified above, and the method of calculating the feature amount is not limited to the first and second examples exemplified above. [Explanation of symbols]

[0061] 1...power supply system, 2...external power source, 3...bus, 4...power distribution line, 5...power load, 6...power feeder, 7...power generation equipment, 9,10,13...switch, 11,14...transformer, 12...customer premises, 20...anomaly detection device, 21...current measurement unit, 22...voltage measurement unit, 23...anomaly detection unit, 24...feature calculation unit, 25...judgment unit, 71...independent power source, 72...power conditioning system (PCS), 231...measurement unit, 232...data preprocessing unit, 233...feature extraction unit, 234...identification calculation unit, 235...identification dictionary, 236...control unit, 237...communication unit, 238...output unit, FL...frame, Ft...feature amount, P...plot, T1...power-on timing, T2...short circuit timing, T3...current limit timing, TH...threshold value.

Claims

1. An abnormality detection device provided in a power supply system including a bus connected to a power supply device, a distribution line connected to the bus, and a power load connected to the distribution line, a current measuring unit that measures a current flowing through the power distribution line; a feature calculation unit that calculates a current feature amount from each of a plurality of frames having different start timings in a first time waveform that is a time waveform of a numerical value related to a current measured by the current measurement unit; a determination unit that determines an abnormality in the power distribution line based on the current feature amount; Equipped with the current feature amount is a numerical value that changes according to the first time waveform and increases or decreases when the first time waveform indicates an abnormality in the power distribution line; The determination unit determines an abnormality in the power distribution line based on a magnitude relationship between the current feature amount calculated by the feature calculation unit and a first threshold.

2. 2. The abnormality detection device according to claim 1, wherein the numerical value relating to the current is at least one numerical value selected from the group consisting of a current value, a frequency, a phase, and a rate of change of the current value of the current measured by the current measuring unit.

3. The abnormality detection device according to claim 1 , wherein the abnormality is a short circuit or a ground fault in the power distribution line.

4. 4. The anomaly detection device according to claim 1, wherein the feature calculation unit obtains numerical values ​​related to the current at each of N timings included in each frame to generate a sequence of N terms, updates the sequence by adding or subtracting adjacent terms included in the sequence, and thereafter repeats the addition or subtraction of adjacent terms to obtain the sum of all terms of the finally generated sequence as the current feature.

5. The feature calculation unit stores in advance an adjustment sequence including N adjustment coefficients; 4. The anomaly detection device according to claim 1, wherein the feature calculation unit obtains numerical values ​​related to the current at each of N timings included in each frame to generate a sequence of N terms, multiplies each of the numerical values ​​of the N terms included in the sequence by an adjustment coefficient for the N terms, and then updates the sequence by adding or subtracting adjacent terms, and thereafter repeats the multiplication of each of the adjustment coefficients for the N terms and the addition or subtraction of adjacent terms, and finally determines the sum of all terms in the sequence generated as the current feature amount.

6. 4. The abnormality detection device according to claim 1, further comprising a line interruption unit that interrupts the power distribution line when the abnormality is detected by the determination unit.

7. 4. The abnormality detection device according to claim 1, wherein a portion including a start timing of one of the plurality of frames and a portion including an end timing of the other of the adjacent frames overlap in time.

8. Further comprising a voltage measurement unit for measuring a voltage in the power distribution line, the feature calculation unit further calculates a voltage feature amount from each of a plurality of frames having different start timings in a second time waveform that is a time waveform of a numerical value related to the voltage measured by the voltage measurement unit; the voltage feature amount is a numerical value that changes according to the second time waveform and increases or decreases when the second time waveform indicates an abnormality in the power distribution line; The abnormality detection device according to any one of claims 1 to 3, wherein the determination unit determines an abnormality in the power distribution line based further on a magnitude relationship between the voltage feature calculated by the feature calculation unit and a second threshold value.

9. A method for detecting an abnormality in a power supply system including a bus connected to a power supply device, a distribution line connected to the bus, and a power load connected to the distribution line, comprising: a current measuring step of measuring a current flowing through the power distribution line; a feature calculation step of calculating a current feature amount from each of a plurality of frames having different start timings in a first time waveform that is a time waveform of a numerical value related to the current measured in the current measurement step; a determination step of determining an abnormality in the power distribution line based on the current feature amount; Including, the current feature amount is a numerical value that changes according to the first time waveform and increases or decreases when the first time waveform indicates an abnormality in the power distribution line; The method for detecting an anomaly in a power supply system, wherein the determining step determines whether an anomaly exists in the power distribution line based on a magnitude relationship between the current feature amount calculated in the feature calculation step and a first threshold value.

Images