Power distribution network fault detection method based on generalized S transformation containing variable factors

Abstract

The invention relates to a power distribution network fault detection method based on generalized S transformation containing variable factors, and belongs to the technical field of power distribution network fault detection. Data such as a time-frequency spectrum and transient energy are obtained through generalized S transformation containing variable factors, and three line selection criteria of a time-frequency spectrum comprehensive similarity coefficient, a transient energy peak value and a transient energy relative entropy are constructed; and multi-criterion fusion is realized in combination with a D-S evidence theory, so that a line selection result of a single criterion is subjected to comprehensive analysis of multi-criterion fusion, and a more accurate line selection result is obtained. According to the method, accurate line selection can be carried out in different fault distances, fault lines, fault initial angles, fault grounding resistances and noise environments, and the method has good adaptability, sensitivity and anti-interference performance. According to the method, the calculated transient energy characteristics are converted into a reliability distribution function to construct an evidence combination, and different fault characteristic quantities are utilized to measure the credibility to construct a line selection criterion, so that mutual complementation and mutual fusion are realized, and a more accurate line selection result is obtained.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network fault detection, and in particular to a distribution network fault detection method based on a generalized S transform containing variable factors. Background Art

[0002] In the power system, we call the whole composed of substations of various voltage levels and transmission and distribution lines as the power grid. It includes three units: substation, transmission and distribution. Among them, the distribution network is a power grid with a voltage level of 35kV and below. It is a very critical and important link that directly faces users and connects substation units with power consumption units. It plays an important role in the distribution of electric energy. At the same time, the distribution network has the characteristics of multiple voltage levels, complex network structure, diverse equipment types, many and wide operating points, and relatively poor safety environment. Therefore, there are relatively more safety risk factors in the distribution network. Relevant data statistics show that more than 95% of power outages in the power grid are mostly caused by faults in the distribution network.

[0003] The neutral point grounding methods of the distribution network are divided into the neutral point grounding method of large current (i.e. large current grounding method) and the neutral point grounding method of small current (i.e. small current grounding method). Among them, the arc generated by the large current grounding method cannot be extinguished automatically when a single-phase grounding fault occurs, and the circuit breaker must trip immediately. On the contrary, the arc generated by the small current grounding method can be extinguished automatically when a single-phase grounding fault occurs. Even if the circuit breaker does not trip immediately, it will not cause the same harm as the large current grounding method.

[0004] Single-phase grounding faults account for about 80% of all faults in the small current grounding mode. When a single-phase grounding fault occurs in a small current grounding system, the capacitance voltage of the two phases that have not experienced a grounding fault will increase to the line voltage, and the symmetry of the line voltage is not broken. When intermittent arc grounding occurs in the distribution network system, as the arc is struck, the arc is extinguished, and the arc is reignited, it is easy to cause arc grounding overvoltage, damage the insulation, and thus expand from a grounding fault to a phase-to-phase short circuit. At the same time, with the increasing scale of the distribution network, the number of feeders is also increasing, and the capacitance current is also increasing, which is easy to cause a single-point fault to expand to two-point or even multi-point faults, posing a huge threat to the safe and stable operation of the power system. According to the latest "Technical Guidelines for Distribution Networks", the technical principles for handling single-phase grounding faults in small current grounding systems have been improved. After a single-phase grounding fault occurs, the original "allowing the power grid to continue to operate for 2 hours" is changed to the current "selective tripping". Therefore, according to the requirements of the latest guidelines, it is most important to find the faulty line and the location of the fault point as soon as possible so that the responsible personnel for maintenance and operation can take effective measures to deal with the fault in a timely manner. Summary of the invention

[0005] The present invention aims to solve the above technical problems and provides a distribution network fault detection method based on generalized S transform with variable factors.

[0006] In order to solve the above technical problems, the technical solution provided by the present invention is:

[0007] A distribution network fault detection method based on generalized S transform with variable factors comprises the following steps:

[0008] Step 1:

[0009] First, the voltage sensor on the bus monitors whether the zero-sequence voltage of the system exceeds 0.05-0.15 times the rated voltage of the bus;

[0010] Step 2:

[0011] The transient zero-sequence current of 1 / 4 or 1 / 2 cycle after the fault is recorded and collected through the current sensors on each feeder;

[0012] Step 3:

[0013] By performing generalized S transform with variable factors on the transient zero-sequence current on each feeder, a clear transient characteristic quantity in the time-frequency domain is obtained, and then the transient signal in the time-frequency domain is subjected to data filtering and denoising based on generalized S transform with variable factors and three-phase synchronization error removal.

[0014] Step 4:

[0015] Calculate the value of the corresponding criterion feature quantity according to the time-frequency spectrum comprehensive similarity coefficient line selection method, transient energy peak line selection method, and transient energy relative entropy line selection method;

[0016] Step 5:

[0017] According to the characteristic values ​​of each criterion, the corresponding fault criterion weight and bus criterion factor are calculated to determine the mass function evidence value of each criterion, and then the normalization constant K and the fused trust function value are calculated according to the DS evidence theory;

[0018] Step 6:

[0019] The line corresponding to the maximum value of the fused trust function is judged as the faulty line, and the line selection is completed.

[0020] Preferably, the time-frequency spectrum comprehensive similarity coefficient line selection method is to judge the fault line by forming a comprehensive similarity coefficient through superposition of the similarity coefficient and the distance coefficient.

[0021] Preferably, the formula of the similarity coefficient is as follows:

[0022]

[0023] Among them, |S ab |≤1, is used to measure the similarity between two lines in the time-frequency domain. K represents the total number of frequency intervals, and N represents the total number of time intervals. The smaller the Sab, the smaller the similarity between the two samples, and the larger the Sab, the greater the similarity. When Sab=1, the two samples are basically consistent, and the similarity coefficient Sab contains positive and negative numbers. The current direction of the fault line is opposite to that of the non-fault line. The similarity coefficient between the fault line and the non-fault line is often opposite to the similarity coefficient between the non-fault lines.

[0024] Preferably, the formula of the distance coefficient is as follows:

[0025]

[0026] The distance coefficient is normalized by the standard, and the formula is as follows:

[0027]

[0028] In the formula, R ab Represents the relative Euclidean distance, D max Represents the maximum Euclidean distance between samples, which can be expressed as

[0029]

[0030] In the formula, GST max (k,n) and GST min (k,n) represent the samples with overall larger values ​​and the samples with overall smaller values ​​in the generalized S matrix respectively;

[0031] A more obvious similarity comparison between the faulty line and the non-faulty line is made in the time-frequency domain. The formula is as follows:

[0032]

[0033] The comprehensive similarity coefficient matrix is ​​normalized, and the formula is as follows:

[0034]

[0035] Where L represents the total number of lines. It is expressed as the sum of the comprehensive similarity coefficients between line i and other lines, excluding line i itself.

[0036] Preferably, the formula of the transient energy peak line selection method is as follows:

[0037]

[0038] Preferably, the transient energy relative entropy line selection method is to sum the transient energies of all lines at each frequency point obtained by calculation to obtain the sum of the transient energies of all lines at each frequency point. The formula is as follows:

[0039]

[0040] The weights of the transient energy of all lines at each frequency point are as follows:

[0041]

[0042] The relative entropy weights of all lines and Combining equation (9) with equation (10), we can get the relative entropy of transient energy between line i and line j as shown in equation (11):

[0043]

[0044] The transient energy comprehensive relative entropy of line i can be obtained by summing the relative entropies between line i and all other lines:

[0045]

[0046] Calculate the transient energy comprehensive relative entropy M of all lines i (i=1,2,…,L), then select the three lines with the largest transient energy comprehensive relative entropy, and sort the three transient energy comprehensive relative entropies from large to small, and record the energy comprehensive relative entropies of the three lines as M i1 、M i2 、M i3 , the specific fault criteria are as follows:

[0047] (1) When the transient energy comprehensive relative entropy corresponding to the three lines satisfies M i1 >M i2 +M i3 When , line i is the fault line;

[0048] (2) When the above inequality is not satisfied, it is determined to be a bus fault.

[0049] After adopting the above method, the present invention has the following advantages:

[0050] The present invention obtains time-frequency spectrum and transient energy data through generalized S transform with variable factors, and constructs three line selection criteria including time-frequency spectrum comprehensive similarity coefficient, transient energy peak value and transient energy relative entropy. Then, the DS evidence theory is combined to realize multi-criteria fusion, so that the line selection result of a single criterion is subjected to comprehensive analysis of multi-criteria fusion to obtain more accurate line selection results.

[0051] The multi-criteria fusion fault line selection method based on the generalized S transform containing variable factors and DS evidence theory can accurately select lines under different fault distances, fault lines, fault initial angles, fault grounding resistances and noise environments, and has good adaptability, sensitivity and anti-interference.

[0052] The present invention converts the time-frequency domain data obtained based on the generalized S transform with variable factors and the calculated transient energy characteristics into a credibility distribution function to construct an evidence combination, and uses different fault characteristic quantities to measure the degree of trust to construct a line selection criterion, thereby complementing and integrating each other to obtain a more accurate line selection result.

[0053] The above summary is for illustrative purposes only and is not intended to be limiting in any way. In addition to the illustrative aspects, embodiments and features described above, further aspects, embodiments and features of the present invention will be readily apparent by reference to the accompanying drawings and the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0055] Figure 1 is a flow chart of the present invention;

[0056] Figure 2 1 is the energy distribution diagram of the feeder of the present invention at each frequency point ((a) sound line 133 feeder; (b) faulty line 134 feeder). DETAILED DESCRIPTION

[0057] Specific embodiments of the present invention will now be mentioned in detail. Although the present invention is described in conjunction with these specific embodiments, it should be appreciated that it is not intended to limit the present invention to these specific embodiments. On the contrary, these embodiments are intended to cover substitutions, changes or equivalent embodiments that may be included in the spirit and scope of the invention defined by the claims. In the following description, a large number of specific details are set forth in order to provide a comprehensive understanding of the present invention. The present invention can be implemented without some or all of these specific details. In other cases, in order not to make the present invention unnecessarily obscure, well-known process operations are not described in detail.

[0058] When used in conjunction with "including", "methods including", or similar language in this specification and the appended claims, the singular forms "a", "an", and "the" include plural references unless the context clearly dictates otherwise. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0059] The present invention is further described in detail below in conjunction with the full text.

[0060] Combined with Figure 1 and Figure 2 , a distribution network fault detection method based on generalized S transform with variable factors, comprising the following steps:

[0061] Step 1:

[0062] First, the voltage sensor on the bus monitors whether the zero-sequence voltage of the system exceeds 0.05-0.15 times the rated voltage of the bus. Specifically, the data sampling object can be achieved 0.5T-1.5T after the fault.

[0063] Step 2:

[0064] The transient zero-sequence current of 1 / 4 or 1 / 2 cycle after the fault is recorded and collected through the current sensors on each feeder;

[0065] Step 3:

[0066] By performing generalized S transform with variable factors on the transient zero-sequence current on each feeder, a clear transient characteristic quantity in the time-frequency domain is obtained, and then the transient signal in the time-frequency domain is subjected to data filtering and denoising based on generalized S transform with variable factors and three-phase synchronization error removal.

[0067] Step 4:

[0068] Calculate the value of the corresponding criterion feature quantity according to the time-frequency spectrum comprehensive similarity coefficient line selection method, transient energy peak line selection method, and transient energy relative entropy line selection method;

[0069] Step 5:

[0070] According to the characteristic values ​​of each criterion, the corresponding fault criterion weight and bus criterion factor are calculated to determine the mass function evidence value of each criterion, and then the normalization constant K and the fused trust function value are calculated according to the DS evidence theory;

[0071] Step 6:

[0072] The line corresponding to the maximum value of the fused trust function is judged as the faulty line, and the line selection is completed.

[0073] The time-frequency spectrum comprehensive similarity coefficient line selection method is to judge the fault line by forming a comprehensive similarity coefficient through superposition of the similarity coefficient and the distance coefficient.

[0074] The formula of the similarity coefficient is as follows:

[0075]

[0076] Among them, |S ab |≤1, is used to measure the similarity between two lines in the time-frequency domain. K represents the total number of frequency intervals, and N represents the total number of time intervals. The smaller the Sab, the smaller the similarity between the two samples, and the larger the Sab, the greater the similarity. When Sab=1, the two samples are basically consistent, and the similarity coefficient Sab contains positive and negative numbers. The current direction of the fault line is opposite to that of the non-fault line. The similarity coefficient between the fault line and the non-fault line is often opposite to the similarity coefficient between the non-fault lines.

[0077] The formula of the distance coefficient is as follows:

[0078]

[0079] The distance coefficient is normalized by the standard, and the formula is as follows:

[0080]

[0081] In the formula, R ab Represents the relative Euclidean distance, D max Represents the maximum Euclidean distance between samples, which can be expressed as

[0082]

[0083] In the formula, GST max (k,n) and GST min (k,n) represent the samples with overall larger values ​​and the samples with overall smaller values ​​in the generalized S matrix respectively;

[0084] A more obvious similarity comparison between the faulty line and the non-faulty line is made in the time-frequency domain. The formula is as follows:

[0085]

[0086] The comprehensive similarity coefficient matrix is ​​normalized, and the formula is as follows:

[0087]

[0088] Where L represents the total number of lines. It is expressed as the sum of the comprehensive similarity coefficients between line i and other lines, excluding line i itself.

[0089] The formula of the transient energy peak line selection method is as follows:

[0090]

[0091] The transient energy relative entropy line selection method is to sum the transient energy of each frequency point of each line obtained by calculation for all lines, and obtain the sum of the transient energy of all lines at each frequency point. The formula is as follows:

[0092]

[0093] The weights of the transient energy of all lines at each frequency point are as follows:

[0094]

[0095] The relative entropy weights of all lines and Combining equation (9) with equation (10), we can get the relative entropy of transient energy between line i and line j as shown in equation (11):

[0096]

[0097] The transient energy comprehensive relative entropy of line i can be obtained by summing the relative entropies between line i and all other lines:

[0098]

[0099] Calculate the transient energy comprehensive relative entropy M of all lines i (i=1,2,…,L), then select the three lines with the largest transient energy comprehensive relative entropy, and sort the three transient energy comprehensive relative entropies from large to small, and record the energy comprehensive relative entropies of the three lines as M i1 、M i2 、M i3 , the specific fault criteria are as follows:

[0100] (1) When the transient energy comprehensive relative entropy corresponding to the three lines satisfies M i1 >M i2 +M i3 When , line i is the fault line;

[0101] (2) When the above inequality is not satisfied, it is determined to be a bus fault.

[0102] When the present invention is implemented, specifically, the time-frequency spectrum comprehensive similarity coefficient line selection method is as follows:

[0103] Similarity measurement is a measure of the degree of similarity between two samples. The more similar the samples are, the greater their similarity is, and the more different the samples are, the smaller their similarity is. Common similarity criteria include correlation distance coefficient (i.e. correlation coefficient), similarity coefficient, distance coefficient (mainly Euclidean distance), similarity distance, etc., among which the correlation coefficient mainly describes the degree of shape similarity between two samples, and the similarity coefficient mainly describes the degree of value similarity between two samples, including the degree of linear correlation of the samples and the high and low positive and negative characteristic differences of the samples themselves. Here, calculating the similarity coefficient between samples mainly highlights the value similarity of the samples, and combines the advantage of the intuitiveness of the distance coefficient to form a comprehensive similarity coefficient to compare the similarity of multiple lines, thereby screening out faulty lines.

[0104] Here we introduce the generalized S-transformation matrix GST(k,n) with variable factors to introduce the similarity coefficient, distance coefficient and comprehensive similarity coefficient respectively. Let GST a (k,n) and GST b (k,n) are the time-frequency spectrum matrices of transient zero-sequence current of line a and line b obtained by generalized S transform.

[0105]

[0106] Where A(k,n) is the amplitude matrix obtained after GST modulus, recorded as the modulus matrix of GST, is the phase angle matrix of GST. It can be seen that the GST matrix can better reflect the time-frequency characteristics of the signal. If it is represented by a time-frequency characteristic diagram, it will be more intuitive and easier to understand than the wavelet transform. Here, it is particularly pointed out that the actual frequency f n Represented as f n =(f s / N)·n, where f s It is represented as sampling frequency, n is represented as the corresponding frequency sampling point, when n = 0, f n =0, it is the GST of the DC component.

[0107] Similarity coefficient:

[0108]

[0109] Among them, |S ab |≤1, used to measure the similarity between two lines in the time-frequency domain, K represents the total number of frequency intervals, and N represents the total number of time intervals. ab The smaller the value, the smaller the similarity between the two samples. ab The larger the value, the greater the similarity. ab =1, the two samples are basically consistent. And the similarity coefficient S abIt contains positive and negative numbers. Since the current direction of the fault line is opposite to that of the non-fault line, the similarity coefficient between the fault line and the non-fault line is often opposite to the similarity coefficient between the non-fault lines.

[0110] Distance coefficient (Euclidean distance):

[0111] Euclidean distance refers to the true distance between two points in n-dimensional space, or the natural length of a vector. This distance coefficient can clearly and intuitively show the differences between samples.

[0112]

[0113] (3) Comprehensive similarity coefficient

[0114] In order to make the distance coefficient and the similarity coefficient superimposed to form a comprehensive similarity coefficient, the distance coefficient is normalized to form the relative Euclidean distance expression as follows:

[0115]

[0116] In the formula, R ab Represents the relative Euclidean distance, D max Represents the maximum Euclidean distance between samples, which can be expressed as:

[0117]

[0118] In the formula, GST max (k,n) and GST min (k,n) represent the samples with overall larger values ​​and the samples with overall smaller values ​​in the generalized S matrix respectively. ab The larger the value is, the greater the distance between samples is, and the smaller the similarity is. ab The smaller it is, the smaller the distance between samples is. ab When =0, the sample points are basically consistent.

[0119] Finally, define the comprehensive similarity coefficient C ab is the similarity coefficient S ab Similarity coefficient R ab The combined value of the similarity coefficient and the distance coefficient is combined to make a more obvious similarity comparison between the fault line and the non-fault line in the time and frequency domain.

[0120]

[0121] It can be seen from the above formula that due to the R between the fault line and the non-fault line ab Very large, and S ab is a negative number, so here we subtract the two to maximize the fault characteristic of the fault line.ab The larger the value, the higher the difference between samples. ab The smaller the value, the smaller the difference between samples. The transient zero-sequence current is converted to the time-frequency domain by the generalized S transform with variable factors, and the comprehensive similarity coefficient matrix of the generalized S transform matrix is ​​obtained by comprehensive similarity coefficient calculation, so that the frequency information of each signal and its phase-amplitude relationship can be obtained in the time domain, so that more abundant fault feature information of the signal can be obtained in the time-frequency domain, and the time-frequency domain analysis and processing of the transient zero-sequence current can be completed.

[0122] In order to construct a more intuitive and clear line selection criterion, the comprehensive similarity coefficient matrix is ​​normalized here:

[0123]

[0124] Get the comprehensive similarity coefficient k of each line i , where L represents the total number of lines, It is expressed as the sum of the comprehensive similarity coefficients between line i and other lines, excluding line i itself.

[0125] From the above series of formulas, it can be seen that the comprehensive similarity coefficient matrix value between the faulty line and the non-faulty line is significantly greater than the comprehensive similarity coefficient matrix value between other non-faulty lines, that is, the comprehensive similarity coefficient k of the faulty line i It will be significantly greater than the comprehensive similarity coefficient k of other non-fault lines. i In order to facilitate fault line selection, a threshold line selection criterion is constructed here. The specific criterion is as follows:

[0126] (1) When k i ≥k set When (i=1,2,…,L), line i is the fault line;

[0127] (2) When k is satisfied for all lines i i <k set When (i=1,2,…,L), it is determined to be a busbar fault.

[0128] The calculation method of transient energy peak line selection method is as follows:

[0129] At the single-phase grounding fault point, there is a transient inductive current with a slow decay and a small amplitude, and a transient capacitive current with a fast decay and a large amplitude. Therefore, the fault current of small current grounding is mainly composed of transient capacitive current. Therefore, the fault line can be judged by the energy peak of the transient zero-sequence current of the fault line.

[0130] According to the principles of physical electrical energetics, the traditional transient energy formula is:

[0131]

[0132] Where W i (t) is expressed as the transient energy time domain function, u0(t) is expressed as the bus zero sequence voltage time domain function, i 0i (t) is expressed as the zero-sequence current time domain function of each feeder, where i = 1, 2, ..., L, which is the feeder number. Since the busbar zero-sequence voltage of each feeder is the same, the voltage can be omitted during integration. Combined with the definition of the generalized S transform mentioned above, the discrete zero-sequence current signal is discretized by the generalized S transform, and the transient energy of each line at each frequency point can be obtained as follows:

[0133]

[0134] Based on the above principle, the GST matrix calculated by the generalized S transform with variable factors of the fault transient zero-sequence current obtained through fault simulation is substituted into formula (7), so as to obtain the transient energy distribution of each line at each frequency point, and draw the corresponding transient energy distribution diagram. Taking feeder 133 and feeder 134 as examples, Figure 2 shown.

[0135] From the above energy distribution Figure 2 It can be seen that the transient energy peak value of the faulty line feeder 134 is about 50 times that of the healthy line feeder 133. Therefore, the transient energy peak value can be used as a characteristic criterion for fault line selection, and the difference is very obvious, which can better distinguish the faulty line from other healthy lines. Here, the transient energy peak value E of all lines at all frequency points is calculated. imax (i=1,2,…,L), then select the three lines with the largest transient energy peaks, and sort the energy peaks from large to small, and record the energy peaks of the three lines as E i1max 、E i2max 、E i3max , the specific fault criteria are as follows:

[0136] (1) When the transient energy peaks corresponding to the three lines meet |E i1max |>|E i2max |+|E i3max |, line i is the fault line;

[0137] (2) When the above inequality is not satisfied, it is determined to be a bus fault.

[0138] The specific calculation method of transient energy relative entropy line selection method is as follows:

[0139] Information entropy is a measure of the uncertainty of information, and relative entropy is also called Kullback-Leibler divergence or information divergence, which is used to measure the difference between the information of two samples. Suppose there are two samples, P = {P1, P2, ..., P n} and Q={Q1,Q2,…,Q n}, then the relative entropy between the two is:

[0140]

[0141] When the relative entropy D KL When (P||Q)=0, it means that the difference in information entropy between sample P and sample Q is 0, that is, the two coincide. As the relative entropy value increases, the difference between the two increases. Therefore, the fault line can be found by comparing the relative entropy of transient energy between lines.

[0142] First, the transient energy of each line at each frequency point calculated in the previous section is summed for all lines to obtain the total transient energy of all lines at each frequency point:

[0143]

[0144] Then, the weights of the transient energy of all lines at each frequency point are obtained as follows:

[0145]

[0146] It can be seen from the formula that the relative entropy weights and Combining equation (9) with equation (10), the relative entropy of transient energy between line i and line j can be obtained as follows:

[0147]

[0148] Finally, the relative entropy between line i and all other lines (except line i) is summed to obtain the transient energy comprehensive relative entropy of line i:

[0149]

[0150] Similarly, the transient energy comprehensive relative entropy M of all lines is also calculated here i (i=1,2,…,L), then select the three lines with the largest transient energy comprehensive relative entropy, and sort the three transient energy comprehensive relative entropies from large to small, and record the energy comprehensive relative entropies of the three lines as M i1 、M i2 、M i3 , the specific fault criteria are as follows:

[0151] (1) When the transient energy comprehensive relative entropy corresponding to the three lines satisfies M i1 >M i2 +M i3 When , line i is the fault line;

[0152] (2) When the above inequality is not satisfied, it is determined to be a bus fault.

[0153] The multi-criteria fusion algorithm based on DS evidence theory is as follows:

[0154] DS Evidence Theory:

[0155] The DS algorithm originates from the field of information fusion. In the process of information processing, a lot of information is uncertain, and the DS algorithm is a theoretical tool for intelligent processing and data fusion of uncertain information. The concept of trust function is introduced, thereby constructing a complete set of strategies for dealing with uncertainty reasoning problems from the two perspectives of "evidence" and "combination". On the basis of the three line selection criteria proposed in the present invention, the identification framework is confirmed, and the basic probability assignment function (BPA) of all lines in each criterion is calculated, that is, different BPA values ​​corresponding to different evidence theories are obtained, and then the three mass functions are combined according to the Dempster synthesis rule, so as to realize the fusion of the results of the three fault line selection criteria, and obtain the fused trust function value to determine the fault line. Here is a detailed introduction to the basic principles of DS evidence theory. The identification framework defined in DS evidence theory can also be called a hypothesis space, which is expressed as a non-empty set, which contains multiple mutually exclusive events, and all elements constitute a set containing all possibilities. For the fault line selection problem of the small current grounding system studied in the present invention, the corresponding identification framework can be defined as:

[0156] Θ={F0,F1,F2,…,F L} (15);

[0157] In the formula, F i (i=1,2,…,L) indicates that a fault occurs on line i, and F0 indicates that a fault occurs on the busbar.

[0158] In the recognition frame Θ, the BPA function is a 2 Θ →[0,1] mass function, and satisfies:

[0159]

[0160] In the formula, m(F) refers to the trust assigned to all sets F in the recognition framework Θ. The sum of the trusts of all sets in the recognition framework Θ is 1, and the sets not in this recognition framework Θ are The trust level is 0.

[0161] for The Dempster synthesis rules of the three mass functions on the identification framework Θ, i.e., the three fault line selection criteria m1, m2, and m3, are:

[0162]

[0163] In the formula, K is the normalization constant, and the specific calculation is as follows:

[0164]

[0165] Substituting formula (17) into formula (18), we can get the trust function values ​​of the bus and each line, which can be used as the fusion result criterion to judge the faulty line.

[0166] Fault line selection method based on multi-criteria fusion:

[0167] When a single-phase grounding fault occurs in a low-current grounding system, the three fault line selection criteria will be calculated simultaneously to obtain the corresponding criterion values, and then enter the line selection process of DS evidence theory multi-criteria fusion. Here we introduce the specific construction of the mass function of the three criteria adopted by the present invention for evidence theory. First, the fault weights constructed by each criterion are as follows:

[0168]

[0169] In the formula, P1(F i ) is expressed as the comprehensive similarity coefficient weight of line i, P2(F i ) represents the transient energy peak weight of line i, P3(F i ) represents the relative entropy weight of transient energy of line i. In order to more clearly distinguish line and bus faults, the criterion factor of bus fault is introduced as follows:

[0170]

[0171] Where P jmax It represents the maximum value of the fault weights of all lines of criterion j. Let the bus criterion factor be multiplied by the fault weight of each criterion to obtain the mass function value of each criterion of all lines and buses as follows:

[0172]

[0173] In the formula, m j (F0) represents the evidence value of the busbar under criterion j, m j (F i ) represents the evidence value of line i under criterion j.

[0174] The present invention firstly monitors whether the zero-sequence voltage of the system exceeds the limit through the voltage sensor on the busbar. If the zero-sequence voltage of the system exceeds 0.05-0.15 times of the rated voltage of the busbar, the fault line selection process begins;

[0175] (2) The transient zero-sequence current of 1 / 4 or 1 / 2 cycle after the fault is recorded and collected by the current sensors on each feeder;

[0176] (3) By performing generalized S transform with variable factors on the transient zero-sequence current on each feeder, a clear transient characteristic quantity in the time-frequency domain is obtained, and then the transient signal in the time-frequency domain is subjected to data filtering and denoising based on generalized S transform with variable factors and the application of three-phase synchronization error removal;

[0177] (4) Calculate the value of the corresponding criterion feature quantity according to the time-frequency spectrum comprehensive similarity coefficient line selection method, transient energy peak line selection method, and transient energy relative entropy line selection method;

[0178] (5) According to the characteristic values ​​of each criterion, the corresponding fault criterion weight and bus criterion factor are calculated to determine the mass function evidence value of each criterion, and then the normalization constant K and the fused trust function value are calculated according to the DS evidence theory;

[0179] (6) The line with the maximum BPA after fusion is the fault line, and the line selection is completed.

[0180] The present invention obtains time-frequency spectrum and transient energy data through generalized S transform with variable factors, and constructs three line selection criteria including time-frequency spectrum comprehensive similarity coefficient, transient energy peak value and transient energy relative entropy. Then, the DS evidence theory is combined to realize multi-criteria fusion, so that the line selection result of a single criterion is subjected to comprehensive analysis of multi-criteria fusion to obtain more accurate line selection results.

[0181] The present invention and its embodiments are described above, and such description is not restrictive. What is shown in the full text is only one of the embodiments of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A distribution network fault detection method based on generalized S transform with variable factors, characterized in that: The following steps are involved: Step 1: First, the voltage sensor on the bus monitors whether the zero-sequence voltage of the system exceeds 0.05-0.15 times the rated voltage of the bus; Step 2: The transient zero-sequence current of 1 / 4 or 1 / 2 cycle after the fault is recorded and collected through the current sensors on each feeder; Step 3: By performing generalized S transform with variable factors on the transient zero-sequence current on each feeder, a clear transient characteristic quantity in the time-frequency domain is obtained, and then the transient signal in the time-frequency domain is subjected to data filtering and denoising based on generalized S transform with variable factors and three-phase synchronization error removal. Step 4: Calculate the value of the corresponding criterion feature quantity according to the time-frequency spectrum comprehensive similarity coefficient line selection method, transient energy peak line selection method, and transient energy relative entropy line selection method; Step 5: According to the characteristic values ​​of each criterion, the corresponding fault criterion weight and bus criterion factor are calculated to determine the mass function evidence value of each criterion, and then the normalization constant K and the fused trust function value are calculated according to the DS evidence theory; Step 6: The line corresponding to the maximum value of the fused trust function is judged as the faulty line, and the line selection is completed.

2. The distribution network fault detection method based on generalized S transform with variable factors according to claim 1 is characterized in that: The time-frequency spectrum comprehensive similarity coefficient line selection method is to judge the fault line by forming a comprehensive similarity coefficient through superposition of the similarity coefficient and the distance coefficient.

3. The distribution network fault detection method based on generalized S transform with variable factors according to claim 2 is characterized in that: The formula of the similarity coefficient is as follows: Among them, |S ab |≤1, is used to measure the similarity between two lines in the time-frequency domain. K represents the total number of frequency intervals, and N represents the total number of time intervals. The smaller the Sab, the smaller the similarity between the two samples, and the larger the Sab, the greater the similarity. When Sab=1, the two samples are basically consistent, and the similarity coefficient Sab contains positive and negative numbers. The current direction of the fault line is opposite to that of the non-fault line. The similarity coefficient between the fault line and the non-fault line is often opposite to the similarity coefficient between the non-fault lines.

4. The method for detecting distribution network faults based on generalized S transform with variable factors according to claim 2, characterized in that: The formula of the distance coefficient is as follows:

5. The distribution network fault detection method based on generalized S transform with variable factors according to claim 2 is characterized in that: The distance coefficient is normalized by the standard, and the formula is as follows: In the formula, R ab Represents the relative Euclidean distance, D max Represents the maximum Euclidean distance between samples, which can be expressed as In the formula, GST max (k,n) and GST min (k,n) represent the samples with overall larger values ​​and the samples with overall smaller values ​​in the generalized S matrix respectively; A more obvious similarity comparison between the faulty line and the non-faulty line is made in the time-frequency domain. The formula is as follows: The comprehensive similarity coefficient matrix is ​​normalized, and the formula is as follows: Where L represents the total number of lines. It is expressed as the sum of the comprehensive similarity coefficients between line i and other lines, excluding line i itself.

6. The distribution network fault detection method based on generalized S transform with variable factors according to claim 1 is characterized in that: The formula of the transient energy peak line selection method is as follows:

7. The distribution network fault detection method based on generalized S transform with variable factors according to claim 1 is characterized in that: The transient energy relative entropy line selection method is to sum the transient energy of each frequency point of each line obtained by calculation for all lines, and obtain the sum of the transient energy of all lines at each frequency point. The formula is as follows: The weights of the transient energy of all lines at each frequency point are as follows: The relative entropy weights of all lines and Combining equation (9) with equation (10), we can get the relative entropy of transient energy between line i and line j as shown in equation (11): The transient energy comprehensive relative entropy of line i can be obtained by summing the relative entropies between line i and all other lines: Calculate the transient energy comprehensive relative entropy M of all lines i (i=1,2,…,L), then select the three lines with the largest transient energy comprehensive relative entropy, and sort the three transient energy comprehensive relative entropies from large to small, and record the energy comprehensive relative entropies of the three lines as M i1 、M i2 、M i3 , the specific fault criteria are as follows: (1) When the transient energy comprehensive relative entropy corresponding to the three lines satisfies M i1 >M i2 +M i3 When , line i is the fault line; (2) When the above inequality is not satisfied, it is determined to be a bus fault.

Images