A method and system for fault location in flexible DC distribution networks

Abstract

This invention discloses a method and system for fault location in flexible DC distribution networks. The method includes: acquiring current signals of DC-side lines in the flexible DC distribution network after a fault; performing CEEMD decomposition on the current signals to obtain IMF components; calculating the bubble entropy of each IMF component to construct a feature vector for the line, wherein the feature vector is composed of the bubble entropy of each IMF component corresponding to the line; and inputting the feature vector into a fault location model for the flexible DC distribution network based on an SVM model to obtain a location selection result. The input of the fault location model is the feature vector of the line, and the output is a location selection result indicating whether the line is faulty. Preferably, the Black Wolf algorithm is introduced to optimize the penalty factor and kernel parameters. Through the above improvements and optimizations, the technical solution of this invention significantly improves the reliability and accuracy of fault location.

Description

Technical Field

[0001] This invention belongs to the field of DC distribution network technology, specifically relating to a method and system for fault location in flexible DC distribution networks. Background Technology

[0002] With the rapid development of power electronics technology and the large demand for new energy sources, flexible DC distribution networks have become an important development direction for urban power distribution due to their advantages such as large transmission capacity, low loss, and high power supply reliability. However, when a flexible DC distribution network fails, the fault current rises rapidly and the peak value is high. But the power electronic devices have limited ability to withstand inrush current, so it is necessary to quickly and accurately select the faulty feeder in order to isolate the faulty feeder in time and ensure the stable operation of normal lines.

[0003] The limited availability of effective information for fault location after a fault in a flexible DC distribution network leads to poor positioning accuracy. Therefore, how to select effective fault features for location is a pressing issue. Commonly used fault feature extraction methods, such as Empirical Mode Decomposition (EMD), suffer from mode aliasing, affecting the accurate extraction of signal features. Summary of the Invention

[0004] The purpose of this invention is to address the technical problem in existing technologies where the limited effective information provided for fault location after a fault in a flexible DC distribution network leads to poor positioning accuracy. Specifically, it aims to overcome the mode aliasing problem inherent in commonly used fault feature extraction methods such as Empirical Mode Decomposition (EMD), which affects the accurate extraction of signal features. Therefore, this invention provides a fault location method and system for flexible DC distribution networks. This method performs CEEMD decomposition on the current signal to obtain IMF components, then calculates the bubble entropy of the IMF components as a feature vector, which is input into the fault location model of the flexible DC distribution network. Notably, this invention uses bubble entropy, reducing the importance of selecting parameters such as scaling factor and embedding dimension in entropy calculation.

[0005] Therefore, the present invention provides the following technical solution:

[0006] On the one hand, the present invention provides a method for fault location in a flexible DC distribution network, comprising the following steps:

[0007] S1: Collect current signals of DC-side lines after a fault in a flexible DC distribution network;

[0008] S2: Perform CEEMD decomposition on the current signal of the line to obtain the IMF component;

[0009] S4: Calculate the bubble entropy of each IMF component and construct the feature vector of the line. The feature vector is composed of the bubble entropy of each IMF component corresponding to the line.

[0010] S5: Input the feature vector into the flexible DC distribution network fault line selection model to obtain the line selection result. The input of the flexible DC distribution network fault line selection model is the feature vector of the line, and the output is the line selection result of whether the line is faulty.

[0011] Further optionally, the calculation of the bubble entropy is expressed as:

[0012]

[0013] In the formula, BE is the bubble entropy. Here, m represents the second-order Ruili entropy corresponding to the m+1 dimension and the m dimension, where m is the embedding dimension. The second-order Ruili entropy is calculated by reconstructing (N-(m-1)τ) m+1 dimension and m dimension sub-vectors for the phase space of the IMF component, where N is the total number of elements in the IMF component and τ is the delay factor.

[0014] Further, optionally, the process of obtaining the second-order Ruili entropy is as follows:

[0015] For each IMF component, phase space reconstruction is performed to obtain several sub-vectors;

[0016] Here, we assume that the IMF components are represented as: x = {x1, x2, ..., x...} N}, x1, x2, x N Let N represent the 1st, 2nd, and Nth elements respectively. The matrix represented by (N-(m-1)τ) m-dimensional sub-vectors reconstructed from these elements is:

[0017]

[0018] In the formula, X(1), X(2), and X(N-(m-1)τ) represent the 1st, 2nd, 3rd, and (N-(m-1)τ)th m-dimensional sub-vectors of the reconstruction, respectively; x(1), x(1+τ), and x(1+(m-1)τ) represent the 1st, 2nd, and mth elements of sub-vector X(1), respectively; x(2), x(2+τ), and x(2+(m-1)τ) represent the 1st, 2nd, and mth elements of sub-vector X(2), respectively; and {x(N-(m-1)τ), x(N-(m-2)τ), ..., x(N)} represent the 1st, 2nd, and mth elements of sub-vector X(N-(m-1)τ). The first element of each sub-vector is determined by the IMF component, x(1) = x1, x(2) = x2, and the latter element is the derivative of the former element.

[0019] Sort the elements in each reconstructed subvector in ascending order and record the number of element swaps n that occur during the sorting process. v ;

[0020] Draw a histogram based on the number of element swaps, and calculate the probability p using normalization. v The second-order Ruili entropy is calculated according to the following formula;

[0021] The formula for calculating the second-order Ruili entropy for an m-dimensional subvector is as follows:

[0022]

[0023] The second-order Ruili entropy corresponding to dimension m+1 and dimension m is calculated using the method described above.

[0024] Furthermore, the fault diagnosis classification model uses the support vector machine (SVM) model, but its classification performance is affected by the penalty factor c and the kernel parameter g, which leads to getting stuck in local optima. There is currently no effective technical means to solve this problem.

[0025] Further optionally, the flexible DC distribution network fault selection model is constructed based on the SVM optimization model, which is an SVM model after optimizing the penalty factor and kernel parameters using the Black Wolf algorithm;

[0026] The fitness function of the Black Wolf algorithm is the recognition accuracy of the SVM model on the training set samples.

[0027] Further optionally, the mathematical model of the black wolf pack surrounding prey in the black wolf algorithm is expressed as:

[0028]

[0029] In the formula, λ is a random number, taking a value between 0 and 1, and the user-defined parameters j and k satisfy a normal distribution. constant σ j and σ k The definition is as follows:

[0030]

[0031] In the formula, parameter ω is a preset coefficient, and the preferred value of parameter ω is 1.5; Γ is the gamma function.

[0032] In this invention, the Black Wolf algorithm is optimized. It fully considers that the faults in flexible DC distribution networks have different fault conditions and that the fault information has some overlap. This avoids the Black Wolf algorithm from easily getting trapped in local optima during the global search process. Random coefficients are introduced to better simulate random behavior in nature, thereby improving the Black Wolf position update formula and enhancing the algorithm's ability to escape local optima.

[0033] Further optionally, the mathematical model of the black wolf pack surrounding prey in the black wolf algorithm is expressed as:

[0034]

[0035] In the formula, u is the current iteration number, I(u) is the position of the black wolf individual in the u-th iteration, and I1(u+1), I2(u+1), and I3(u+1) represent intermediate variables, which are respectively determined by the optimal position of the black wolf individual in the black wolf population. α (u), Second-best Black Wolf Individual Position I β (u), Second-best Black Wolf Individual Position I γ (u) The position of the moving black wolf individual; A1, A2, and A3 represent the optimal position I of the black wolf individual, respectively. α (u), Second-best Black Wolf Individual Position I β (u), Second-best Black Wolf Individual Position I γ (u) represents the convergence factor, where C1, C2, and C3 represent the optimal position I of the black wolf individual. α (u), Second-best Black Wolf Individual Position I β (u), Second-best Black Wolf Individual Position I γ (u) corresponds to the oscillation factor, I bwo (u+1) represents the updated position of the black wolf individual after adjustment based on the positions of the α wolf, β wolf, and γ wolf.

[0036] Further, optionally, the best individual and the second-best individual in the population of the Black Wolf algorithm can be used as the penalty factor and kernel parameter, respectively.

[0037] Secondly, the present invention provides a route selection system based on the above-mentioned route selection method, comprising:

[0038] The current signal acquisition module is used to acquire the current signal of the DC side line after a fault in the flexible DC distribution network.

[0039] The decomposition module is used to perform CEEMD decomposition on the current signal of the line to obtain IMF components;

[0040] The feature vector construction module is used to calculate the bubble entropy of each IMF component and construct the feature vector of the line. The feature vector is composed of the bubble entropy of each IMF component corresponding to the line.

[0041] The line selection module is used to input the feature vector into a flexible DC distribution network fault line selection model based on the SVM model to obtain the line selection result. The input of the flexible DC distribution network fault line selection model is the feature vector of the line, and the output is the line selection result indicating whether the line is faulty.

[0042] In three aspects, the present invention provides a computer-readable storage medium storing a computer program, which is invoked by a processor to implement:

[0043] Steps of a fault location method for a flexible DC distribution network.

[0044] In four aspects, the present invention provides an electronic terminal, comprising at least:

[0045] One or more processors;

[0046] A memory that stores one or more computer programs;

[0047] The processor calls the computer program to implement:

[0048] Steps of a fault location method for a flexible DC distribution network.

[0049] Beneficial effects

[0050] Compared with existing methods, the advantages of the present invention are:

[0051] The technical solution of this invention performs CEEMD decomposition on the current signal, reducing noise contamination in the signal and solving the mode aliasing problem. In particular, the technical solution of this invention selects bubble entropy as a feature parameter, which fully considers that after a line fault occurs in this application, the fault current will discharge rapidly within a few milliseconds, resulting in too little available fault data and too little fault information. Bubble entropy has low requirements for parameter selection, converges faster than other entropy algorithms, is more stable, and has stronger discrimination ability. It can extract fault information features in a very short time. At the same time, using bubble entropy to extract features improves accuracy and reduces the importance of selecting parameters such as scaling factor and embedding dimension in entropy calculation. Therefore, the technical solution of this invention introduces bubble entropy as a feature parameter for the first time in this application field, which can greatly improve the accuracy and reliability of fault line selection.

[0052] A further preferred embodiment of this invention employs the Black Wolf algorithm to optimize the penalty factor and kernel parameters of the SVM model, preventing the SVM flexible DC distribution network route selection model from getting trapped in local optima and improving the accuracy of its route selection results. Traditional support vector machines struggle to determine the optimal penalty parameters and kernel function; the Black Wolf algorithm, with its simple structure and fewer parameter settings, significantly improves the prediction speed and accuracy of the support vector machine, achieving excellent results in complex search spaces. Simultaneously, it enhances the training speed and robustness of the route selection model, as faults present different conditions, demonstrating strong adaptability to various fault conditions. Attached Figure Description

[0053] Figure 1 It is a schematic flow diagram of the present invention;

[0054] Figure 2 This is a flowchart of the Black Wolf algorithm for optimizing the SVM model in this invention;

[0055] Figure 3 This is a power distribution network simulation diagram of the present invention;

[0056] Figure 4 This is a CEEMD exploded view of the faulty circuit of this invention. Detailed Implementation

[0057] The present invention provides a fault location method for flexible DC distribution networks. Its core is to perform CEEMD decomposition on the current signal to obtain IMF components, and calculate the bubble entropy of each IMF component as a feature vector, which is then input into the fault location model to effectively improve the accuracy and reliability of fault location. Furthermore, the present invention preferably optimizes the Support Vector Machine (SVM) by introducing the Black Wolf algorithm to optimize the penalty factor c and kernel parameter g, thereby improving the classification accuracy of the SVM. Finally, a fault location model for flexible DC distribution networks is constructed based on the optimized SVM model. The present invention will be further described below with reference to embodiments.

[0058] Example 1:

[0059] This embodiment provides a fault location method for a flexible DC distribution network, including the following steps:

[0060] S1: Collect the current signal of the DC side line after a fault in the flexible DC distribution network.

[0061] S2: Perform CEEMD decomposition on the current signal of the line to obtain the IMF component.

[0062] The CEEMD decomposition process is as follows:

[0063] (1) Initialize the noise amplitude and the number of decompositions I = 10, and let the number of decompositions i = 1, indicating the first decomposition. It should be understood that the number of decompositions in this embodiment is for illustrative purposes only. In other feasible embodiments, the number of decompositions can be adaptively adjusted according to actual needs.

[0064] (2) Add a pair of white noises with equal amplitude and opposite signs to the current signal y(t) to obtain a new signal. Then perform EMD decomposition on the new signal to obtain the new signal N. i + (t), N i - (t) is represented as:

[0065] N i + (t)=y(t)+n i (t)

[0066] N i - (t)=y(t)-n i (t)

[0067] Among them, n i (t) represents the white noise sequence added at the i-th time, and N i + (t) and N i - (t) represents the new signal sequence after adding white noise. After EMD decomposition respectively, k IMF components are obtained:

[0068]

[0069] Among them, c+j,i and c-j,i represent the j-th IMF component corresponding to the new signal N i + (t), N i - (t) obtained respectively, and k represents the number of IMF components.

[0070] (3) When i < I, let i = i + 1, and add different white noises and repeat step (2);

[0071] (4) Calculate the IMF average value of each IMF component after 10 times of EMD decomposition as the final IMF component:

[0072]

[0073] Among them, c j represents the j-th IMF component obtained after CEEMD decomposition.

[0074] It should be understood that several IMF components can be obtained according to the above method.

[0075] Under different faults, the frequency distribution information in each IMF component is different, that is, the uncertainty information contained in the IMF component is different. And the IMF component is a time series of n segments, and the accuracy of sending it into the fault line selection model is not high. As a method of feature extraction, entropy can improve the accuracy in dealing with information uncertainty and randomness. The traditional entropy algorithm depends on parameters, resulting in a decrease in the sensitivity and stability of the entropy algorithm for characterizing fault signals. The present invention adopts bubble entropy, reducing the importance of parameter selection such as the scale factor and embedding dimension in entropy calculation.

[0076] S4: Calculate the bubble entropy of each IMF component and construct a feature vector of the line, where the feature vector is composed of the bubble entropy of each IMF component corresponding to the line.

[0077] The calculation process of the bubble entropy is as follows:

[0078] (1) Reconstruct the phase space of the time series (each IMF component) x = {x1, x2,..., x N} to obtain a matrix:

[0079]

[0080] In this matrix, each element in a row is the derivative of the previous element. Each row is considered a reconstructed component. When τ is 1, there are N-m+1 reconstructed components. m is the embedding dimension, and τ is the delay factor. In this embodiment, the delay factor is 1. In other feasible embodiments, the value of the delay factor is adaptively adjusted according to requirements and accuracy. The elements x1, x2, ..., x of the original IMF component are... N Used to determine the first element value of a subvector in the reconstructed matrix, with values ​​assigned one-to-one.

[0081] After matrix reconstruction, each row becomes a sub-vector, resulting in N-(m-1)τ m-dimensional sub-vectors X(w):

[0082] X(w)={x(w),x(w+τ),…,x(w+m-τ)}, w=1,2,…,N-(m-1)τ.

[0083] (2) Sort the elements of each row in the reconstructed matrix in ascending order. The number of swaps required to achieve ascending order in each row is uncertain. Count the number of swaps that occur during the sorting process. v represents the number of swaps. For example, if each row has m elements, starting from the first element, compare the following m-1 elements. If the element to the right is greater than the element to the left, its position remains unchanged; otherwise, swap their positions. The maximum number of swaps for each row is:

[0084]

[0085] (3) According to n v Draw a histogram of the values ​​and normalize it to obtain the probability p. v And calculate the second-order Ruili entropy. The calculation formula is as follows:

[0086]

[0087] Where, p v n represents the probability of a swap number of times v; v The number of swaps is v; N represents the length of the IMF component; m represents the dimension of the matrix.

[0088] (4) Repeat steps (1) to (3) to calculate the vector of elements m+1. (The embedding dimension will be set to m+1), and the bubble entropy BE will be calculated:

[0089]

[0090] It should be understood that in this embodiment, the second-order Ruili entropy of the m-dimensional dimension is calculated first, and then the second-order Ruili entropy of the m+1-dimensional dimension is calculated. In other feasible embodiments, there are no special requirements for the calculation order.

[0091] This invention obtains the bubble entropy corresponding to each IMF component, uses the bubble entropy value as a fault feature vector, and inputs the feature vector into a flexible DC distribution network fault selection model for training. Finally, the trained model is tested using test set samples. The training set is constructed by using the bubble entropy of the line's current signal as the feature vector of the sample, and the fault code as the label.

[0092] S5: Input the feature vector into the SVM-based flexible DC distribution network fault selection model to obtain the selection result. The input of the flexible DC distribution network fault selection model is the feature vector of the line, and the output is the selection result of whether the line is faulty.

[0093] In this embodiment, the SVM model has been optimized. In other feasible embodiments, the technical solution of the present invention, based on the introduction of bubble entropy, also uses the existing SVM classification model to meet the basic requirements and improve the line selection accuracy.

[0094] This embodiment introduces the Black Wolf algorithm to optimize the penalty factor and kernel parameters of the SVM model. The penalty factor and kernel parameters of the SVM model are existing parameters, and their application in SVM models is also existing technology; therefore, they will not be described in detail, but only briefly. The focus of this invention is to introduce the Black Wolf algorithm to optimize the penalty factor and kernel parameters.

[0095] (1) Assume the training set samples to be classified linearly are:

[0096] X = {(a1,b1),(a2,b2),…,(a...} l ,b l )}

[0097] Among them, a i For the data sample, i.e., the bubble entropy of the line current signal, satisfying a i ∈R n , R n Let bi represent an n-dimensional real space, where bi is the type label, satisfying b i ∈{-1,1}, i=1,2,…,l, a value of -1 represents a faulty line, and a value of 1 represents a normal line; the SVM classification model can be understood as a i The input is a function b = xa + y that finds a and b, used to infer the value of b for any a. x is the normal vector of the optimal hyperplane, and y is the bias.

[0098] Introducing slack variable ξ into the SVM modeli The optimization problem of the SVM model becomes:

[0099]

[0100] Wherein, slack variable ξ i ≥0, c is the penalty factor. The penalty factor determines the degree of punishment for misclassified samples by SVM. When the value is large, the generalization ability of SVM becomes weaker. When the value is small, the fitting error will increase, which will bring greater empirical risk.

[0101] Computation in high-dimensional spaces is complex, therefore a kernel function is introduced. This invention uses the radial basis function kernel as the kernel function of SVM, and its expression is as follows:

[0102] K(a,b)=exp(-ga-b 2 )

[0103] Among them, the kernel parameter g>0. The value of the kernel parameter affects the classification performance of the SVM model. When the value is large, the SVM cannot effectively identify and classify different samples. When the value is small, the accuracy of the SVM in identifying unknown samples is not high.

[0104] Since the penalty factor and kernel function are both existing technologies in the SVM model, and the SVM model itself is also existing technology, they will not be described in detail. The purpose of this invention is to introduce the Black Wolf algorithm to optimize the penalty factor and kernel function of the SVM model, so that when applied in this field, it can effectively improve the accuracy and reliability of fault line selection.

[0105] Regarding the Black Wolf Algorithm:

[0106] Let the spatial dimension be d, and the black wolf population size be S. Let the position of the q-th black wolf be I. q Expressed as:

[0107]

[0108] Let α denote the best individual in the population, β denote the second best individual, γ denote the next second best individual, and δ denote the others. Then, the mathematical model of a black wolf pack surrounding its prey can be expressed as:

[0109] I(u+1)=I tg (u)-A·|C·I tg (u)-I(u)

[0110] Where u is the current iteration number, I tg (u) represents the prey's position, and I(u) and I(u+1) represent the positions of the black wolf before and after its movement in the u-th and u+1-th iterations, respectively. A is the convergence factor, and C is the oscillation factor, calculated as follows:

[0111] A = 2a·r1-a, C = 2·r2

[0112] Where r1 and r2 are both random numbers between [0,1]; a is a control parameter that increases linearly with the number of iterations.

[0113] Subtract, the calculation is as follows:

[0114]

[0115] Among them, U max To determine the maximum number of iterations, a in and a fin These represent the initial and final values ​​of the control parameter 'a', respectively. During the hunt, the α, β, and γ wolves are closest to the prey, while the other black wolves follow the alpha wolf around the prey. Position the α, β, and γ wolves I... α (u), I β (u) and I γ (u) is considered the prey location I tg (u) Substituting into the mathematical model of a pack of black wolves surrounding their prey, we get:

[0116]

[0117] In the formula, I1(u+1), I2(u+1), and I3(u+1) represent intermediate variables, which are determined based on the optimal position I of the black wolf individual in the black wolf population. α (u), Second-best Black Wolf Individual Position I β (u), Second-best Black Wolf Individual Position I γ (u) The position of the moving black wolf individual; A1, A2, and A3 represent the optimal position I of the black wolf individual, respectively. α (u), Second-best Black Wolf Individual Position I β (u), Second-best Black Wolf Individual Position I γ (u) represents the convergence factor, where C1, C2, and C3 represent the optimal position I of the black wolf individual. α (u), Second-best Black Wolf Individual Position I β (u), Second-best Black Wolf Individual Position I γ (u) corresponds to the oscillation factor, and Ibwo(u+1) represents the updated position of the black wolf individual after adjustment based on the positions of the α, β, and γ wolves. After iteration, I α (u) represents the location of the prey that the black wolf pack eventually found.

[0118] It should be noted that this embodiment preferably optimizes the above-mentioned black wolf position update formula as follows:

[0119]

[0120] Where λ is a random number, taking values ​​between 0 and 1, and j and k follow a normal distribution.

[0121] constant σ j and σ k The definition is as follows:

[0122]

[0123] In the formula, the parameter ω takes the value of 1.5.

[0124] Based on the above theoretical statements, the specific steps of the Black Wolf algorithm to optimize the SVM model are as follows:

[0125] (1) Initialize algorithm parameters: population size S, maximum number of iterations U max The initial value of control parameter a in and the final value a fin The range of values ​​for the penalty factor c and the kernel parameter g to be optimized;

[0126] (2) Based on the range of values ​​of the parameters to be optimized, a two-dimensional initial black wolf population {I} is randomly generated. q ,q=1,2,…,S;

[0127] (3) Put I α and I β An SVM model is established using I as the penalty factor c and kernel parameter g, respectively. It should be understood that the penalty factor and kernel parameter are of the same dimension; therefore, it is preferable to use I as the penalty factor c and kernel parameter g, respectively. α and I β The penalty factor c and kernel parameter g are used for optimization, respectively. In other feasible embodiments, the penalty factor and kernel parameter can be constructed into a parameter vector to represent the position of the black wolf, and then the optimal position of the black wolf can be found as the final determined penalty factor and kernel parameter.

[0128] (4) Using the recognition accuracy of SVM on the training set samples as the fitness function, calculate the fitness value f(I) of all black wolf individuals. q );

[0129] (5) Select the top three individuals with the highest fitness in ascending order and label them α wolf, β wolf, and γ wolf. Their positions are denoted as I. α I β and I γ ;

[0130] (6) Calculate I bwo And update the location of the black wolf pack I q ;

[0131] (7) Recalculate the fitness value f(I) of all individual black wolves. q), and update the α wolf, β wolf, and γ wolf and their positions I. α , I β and I γ ;

[0132] (8) If u reaches the maximum number of iterations U max If so, the algorithm terminates and I is set to 1. α and I β As the optimal penalty factor c best and kernel parameter g best Output the value; otherwise, u = u + 1, and return to step (6).

[0133] (9) Use the optimal penalty factor c best and kernel parameter g best The training set samples are used to train and learn, resulting in an SVM fault selection model optimized by the Black Wolf algorithm.

[0134] (10) Input the test set samples into the trained SVM fault route selection model optimized by the Black Wolf algorithm to verify the route selection results.

[0135] Application examples:

[0136] This invention uses Matlab / Simulink to build such a system. Figure 3 The system shown is a 110kV substation with 4 feedback lines on the 10kV side, and a sampling frequency of 100kHz. The parameters are set as follows: number of IMFs after EMD decomposition k = 5; embedding dimension m = 6; delay factor τ = 1; the feature vector based on bubble entropy is represented as [e1,e2,e3,e4]; population size S = 20; maximum number of iterations U. max =200, the penalty factor c and the kernel parameter g both range from [0.01, 100], and the initial value of the control parameter a is a. in =2 and the final value a fin =0.

[0137] The above model was configured with four different grounding fault types: metallic grounding, grounding resistance of 10Ω, 100Ω, and high-resistance grounding of 1500Ω. The initial fault angles were set to 0°, 30°, 45°, 60°, and 90° (a total of 5 angles). The fault location from the busbar was set to 10%, 30%, 50%, 70%, and 90% (a total of 5 positions). Four outgoing lines yielded 4×4×5×5 = 400 sets of fault data. To ensure the reliability of the line selection results, 90 sets of fault data from each outgoing line were selected as the training set, and the remaining 10 sets were used as the test set. A fault on outgoing line 1 was classified as the first type, and so on, with a fault on outgoing line 4 being classified as the fourth type. Simulation results show that by setting different fault resistances, initial fault angles, and fault location from the busbar, the line selection accuracy reached 98%.

[0138] Example 2:

[0139] This embodiment also provides a fault location system based on a flexible DC distribution network fault location method, including: a current signal acquisition module, a decomposition module, a feature vector construction module, and a fault location module.

[0140] Among them, the current signal acquisition module is used to acquire the current signal of the DC side line after a fault in the flexible DC distribution network;

[0141] The decomposition module is used to perform CEEMD decomposition on the current signal of the line to obtain IMF components;

[0142] The feature vector construction module is used to calculate the bubble entropy of each IMF component and construct the feature vector of the line. The feature vector is composed of the bubble entropy of each IMF component corresponding to the line.

[0143] The fault location module is used to input the feature vector into the flexible DC distribution network fault location model to obtain the fault location result. The input of the flexible DC distribution network fault location model is the feature vector of the line, and the output is the fault location result of the line. Preferably, the flexible DC distribution network fault location model is based on the SVM model, and even more preferably, the Black Wolf algorithm is introduced to optimize the penalty factor and kernel parameters of the flexible SVM model.

[0144] It should be understood that the specific implementation process of each module is described in the above method. This invention will not repeat the details here. The above division of functional modules is only for illustrative purposes. In some embodiments, some functional modules can be combined and some functional modules can be separated. Each functional module can be implemented in software, hardware, or a combination of software and hardware. The software and hardware devices include, but are not limited to, general-purpose computer equipment, programmable gate arrays, digital signal processors, microprocessors and their corresponding programming or burning software.

[0145] Example 3:

[0146] This embodiment also provides a computer-readable storage medium storing a computer program, which is called by a processor to implement: a fault location method based on a flexible DC distribution network. Specifically, the method is as follows:

[0147] S1: Collect current signals of DC-side lines after a fault in a flexible DC distribution network;

[0148] S2: Perform CEEMD decomposition on the current signal of the line to obtain the IMF component;

[0149] S4: Calculate the bubble entropy of each IMF component and construct the feature vector of the line. The feature vector is composed of the bubble entropy of each IMF component corresponding to the line.

[0150] S5: Input the feature vector into the flexible DC distribution network fault line selection model to obtain the line selection result. The input of the flexible DC distribution network fault line selection model is the feature vector of the line, and the output is the line selection result of whether the line is faulty.

[0151] Among them, the preferred flexible DC distribution network fault location model is based on the SVM model, and the preferred model introduces the Black Wolf algorithm to optimize the penalty factor and kernel parameters of the flexible SVM model.

[0152] Example 4:

[0153] This embodiment also provides an electronic terminal, comprising at least: one or more processors; and a memory storing one or more computer programs; wherein the processor calls the computer programs to implement: a fault location method based on a flexible DC distribution network. Specific execution:

[0154] S1: Collect current signals of DC-side lines after a fault in a flexible DC distribution network;

[0155] S2: Perform CEEMD decomposition on the current signal of the line to obtain the IMF component;

[0156] S4: Calculate the bubble entropy of each IMF component and construct the feature vector of the line. The feature vector is composed of the bubble entropy of each IMF component corresponding to the line.

[0157] S5: Input the feature vector into the flexible DC distribution network fault line selection model to obtain the line selection result. The input of the flexible DC distribution network fault line selection model is the feature vector of the line, and the output is the line selection result of whether the line is faulty.

[0158] Among them, the preferred flexible DC distribution network fault location model is based on the SVM model, and the preferred model introduces the Black Wolf algorithm to optimize the penalty factor and kernel parameters of the flexible SVM model.

[0159] Please refer to the explanation of the method above for the specific implementation process of each step.

[0160] It should be understood that, in the embodiments of the present invention, the processor may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor. The memory may include read-only memory and random access memory, and provides instructions and data to the processor. A portion of the memory may also include non-volatile random access memory. For example, the memory may also store device type information.

[0161] The readable storage medium is a computer-readable storage medium, which can be an internal storage unit of the hardware and software device described in any of the foregoing embodiments, such as the hard drive or memory of the controller. The readable storage medium can also be an external storage device of the controller, such as a plug-in hard drive, Smart MediaCard (SMC), Secure Digital (SD) card, or Flash Card equipped on the controller. Further, the readable storage medium can include both internal storage units and external storage devices of the controller. The readable storage medium is used to store the computer program and other programs and data required by the controller. The readable storage medium can also be used to temporarily store data that has been output or will be output.

[0162] Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned readable storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0163] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-readable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. This application refers to flowchart illustrations and / or instructions executed by a processor of a method, apparatus (system), and computer program product according to embodiments of this application to create means for implementing the functions specified in one or more flowchart illustrations and / or one or more block diagrams. These computer program instructions may also be stored in a computer-readable storage medium capable of directing a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowchart illustrations and / or one or more block diagrams. These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more blocks of a block diagram.

[0164] It should be emphasized that the examples described in this invention are illustrative rather than limiting. Therefore, this invention is not limited to the examples described in the specific embodiments. Any other embodiments derived by those skilled in the art based on the technical solutions of this invention, without departing from the spirit and scope of this invention, whether modifications or substitutions, are also within the protection scope of this invention.

Claims

1. A method for fault location in a flexible DC distribution network, characterized in that: Includes the following steps: S1: Collect current signals of DC-side lines after a fault in a flexible DC distribution network; S2: Perform CEEMD decomposition on the current signal of the line to obtain the IMF component; S4: Calculate the bubble entropy of each IMF component and construct the feature vector of the line. The feature vector is composed of the bubble entropy of each IMF component corresponding to the line. S5: Input the feature vector into the flexible DC distribution network fault line selection model to obtain the line selection result. The input of the flexible DC distribution network fault line selection model is the feature vector of the line, and the output is the line selection result of whether the line is faulty.

2. The method according to claim 1, characterized in that: The bubble entropy is calculated as follows: In the formula, BE is the bubble entropy. Here, m represents the second-order Ruili entropy corresponding to the m+1 dimension and the m dimension, where m is the embedding dimension. The second-order Ruili entropy is calculated by reconstructing (N-(m-1)τ) m+1 dimension and m dimension sub-vectors for the phase space of the IMF component, where N is the total number of elements in the IMF component and τ is the delay factor.

3. The method according to claim 2, characterized in that: The process of obtaining the second-order Ruili entropy is as follows: For each IMF component, phase space reconstruction is performed to obtain several sub-vectors; Here, we assume that the IMF components are represented as: x = {x1, x2, ..., x...} N }, x1, x2, x N Let N represent the 1st, 2nd, and Nth elements respectively. The matrix represented by (N-(m-1)τ) m-dimensional sub-vectors reconstructed from these elements is: In the formula, X(1), X(2), and X(N-(m-1)τ) represent the 1st, 2nd, 3rd, and (N-(m-1)τ)th m-dimensional sub-vectors of the reconstruction, respectively; x(1), x(1+τ), and x(1+(m-1)τ) represent the 1st, 2nd, and mth elements of sub-vector X(1), respectively; x(2), x(2+τ), and x(2+(m-1)τ) represent the 1st, 2nd, and mth elements of sub-vector X(2), respectively; and {x(N-(m-1)τ), x(N-(m-2)τ), ..., x(N)} represent the 1st, 2nd, and mth elements of sub-vector X(N-(m-1)τ). The first element of each sub-vector is determined by the IMF component, x(1) = x1, x(2) = x2, and the latter element is the derivative of the former element. Sort the elements in each reconstructed subvector in ascending order and record the number of element swaps n that occur during the sorting process. v ; Draw a histogram based on the number of element swaps, and calculate the probability p using normalization. v The second-order Ruili entropy is calculated according to the following formula; The formula for calculating the second-order Ruili entropy for an m-dimensional subvector is as follows: The second-order Ruili entropy corresponding to dimension m+1 and dimension m is calculated using the method described above.

4. The method according to claim 1, characterized in that: The flexible DC distribution network fault selection model is constructed based on the SVM optimization model, which is an SVM model after optimizing the penalty factor and kernel parameters using the Black Wolf algorithm. The fitness function of the Black Wolf algorithm is the recognition accuracy of the SVM model on the training set samples.

5. The method according to claim 4, characterized in that: The mathematical model of the black wolf pack surrounding prey in the black wolf algorithm is expressed as follows: In the formula, λ is a random number, taking a value between 0 and 1, and the user-defined parameters j and k satisfy a normal distribution. constant σ j and σ k The definition is as follows: In the formula, parameter ω is a preset coefficient, and Γ is the gamma function.

6. The method according to claim 4, characterized in that: The mathematical model of the black wolf pack surrounding prey in the black wolf algorithm is expressed as follows: In the formula, u is the current iteration number, I(u) is the position of the black wolf individual in the u-th iteration, and I1(u+1), I2(u+1), and I3(u+1) represent intermediate variables, which are respectively determined based on the optimal position I of the black wolf individual in the black wolf population. α (u), Second-best Black Wolf Individual Position I β (u), Second-best Black Wolf Individual Position I γ (u) The position of the moving black wolf individual; A1, A2, and A3 represent the optimal position I of the black wolf individual, respectively. α (u), Second-best Black Wolf Individual Position I β (u), Second-best Black Wolf Individual Position I γ (u) represents the convergence factor, and C1, C2, and C3 represent the optimal position I of the black wolf individual. α (u), Second-best Black Wolf Individual Position I β (u), Second-best Black Wolf Individual Position I γ (u) corresponds to the oscillation factor, I bwo (u+1) represents the updated position of the black wolf individual after adjustment based on the positions of the α wolf, β wolf, and γ wolf.

7. The method according to claim 4, characterized in that: The best and second-best individuals in the population of the Black Wolf algorithm are used as the penalty factor and kernel parameter, respectively.

8. A system based on the method of any one of claims 1-7, characterized in that: include: The current signal acquisition module is used to acquire the current signal of the DC side line after a fault in the flexible DC distribution network. The decomposition module is used to perform CEEMD decomposition on the current signal of the line to obtain IMF components; The feature vector construction module is used to calculate the bubble entropy of each IMF component and construct the feature vector of the line. The feature vector is composed of the bubble entropy of each IMF component corresponding to the line. The line selection module is used to input the feature vector into the flexible DC distribution network fault line selection model to obtain the line selection result. The input of the flexible DC distribution network fault line selection model is the feature vector of the line, and the output is the line selection result of whether the line is faulty.

9. A computer-readable storage medium, characterized in that: The computer program is stored and is invoked by the processor to implement: The steps of the method according to any one of claims 1-8.

10. An electronic terminal, characterized in that: At least including: One or more processors; A memory that stores one or more computer programs; The processor calls the computer program to implement: The steps of the method according to any one of claims 1-8.

Images