Power transmission line equipment fault classification method, system and equipment based on DWT-RBF

Abstract

The invention discloses a DWT-RBF-based power transmission line equipment fault classification method, system and equipment, and the method comprises the steps: firstly building a simulation system of three-phase series compensation power, defining a fault type, then obtaining a sampling signal of a transient current when a power transmission line has a fault, carrying out the decomposition through DWT, and then carrying out the denoising based on an improved wavelet threshold algorithm, the method comprises the following steps: sampling a DWT-RBF signal, acquiring an energy value of the sampling signal, performing normalization processing on the energy value, finally optimizing a DWT-RBF simulation system based on an IWOA algorithm, and inputting the energy value to complete fault classification; according to the method, the sampling signals are decomposed through DWT, fault features are extracted, the response speed of fault classification is increased, the wavelet threshold denoising algorithm is adopted for noise filtering, the recognition accuracy is improved, meanwhile, the RBF model is optimized through the IWOA algorithm, the convergence speed and generalization fitting capacity of the model are improved, and the fault classification accuracy is improved. Therefore, the method has good network performance, and can accurately and quickly classify different types of power transmission line equipment faults.

Description

Technical Field

[0001] The invention relates to a fault classification method, and belongs to the technical field of power transmission line fault diagnosis, and in particular to a power transmission line equipment fault classification method, system and equipment based on DWT-RBF. Background Art

[0002] As one of the indispensable infrastructures in modern society, the safety and stability of the power system are crucial to ensuring social life and economic development. Power system failures may lead to power outages, equipment damage, and even accidents. Therefore, it is necessary to diagnose and handle failures in a timely and accurate manner to ensure the safe operation of the power system.

[0003] In a power system fault, when two or more conductors are in contact with each other or the ground, the fault occurs as a single-line to ground, double-line, double-line to ground and three-line faults. These faults will cause more serious harm and affect the power system equipment and power quality. However, the fault state is difficult to detect due to its short duration. Therefore, it is necessary to use a high sampling rate transient detection instrument and combine it with a transient signal processing and analysis method to complete the fault classification. In recent years, there have been a variety of studies on the transmission line fault type judgment method based on intelligent theory. However, in the existing methods, the fault type judgment based on the expert system is prone to a "dead loop" of reasoning search, and the accuracy is poor. Although the fault type judgment based on conventional neural networks has a certain fault tolerance capability, the network utilization efficiency is not high, and the network learning and training speed is slow. Summary of the invention

[0004] The purpose of the present invention is to overcome the above-mentioned defects and problems existing in the prior art and to provide a DWT-RBF-based transmission line equipment fault classification method, system and equipment with high accuracy and good performance.

[0005] To achieve the above objectives, the technical solution of the present invention is: a DWT-RBF-based power transmission line equipment fault classification method, comprising:

[0006] S1. Build a DWT-RBF simulation system for three-phase series compensation power;

[0007] S2, obtaining a sampling signal of transient current when a transmission line fails, decomposing the sampling signal using DWT, then denoising based on an improved wavelet threshold algorithm, and obtaining an energy value of the sampling signal;

[0008] S3, normalizing the energy value of the sampled signal, and obtaining the maximum and minimum values ​​of the energy value;

[0009] S4. Optimizing the DWT-RBF simulation system based on the IWOA algorithm; the IWOA algorithm optimization includes: any one or any combination of the optimal point set optimization initialization population, elite reverse learning strategy, nonlinear parameter control, and random difference mutation strategy;

[0010] S5. Based on the optimized DWT-RBF simulation system, the energy value is input into the optimized DWT-RBF neural network to complete the fault classification.

[0011] In step S2, the improved wavelet threshold algorithm denoising specifically includes:

[0012] S21. Based on the normal distribution principle of Gaussian distribution, the sampling signal noise is distributed in the interval of [-3σ, 3σ]. The maximum and minimum thresholds are used to denoise the sampling data of each layer component in the interval of [-3σ, 3σ]. The denoising expression is as follows:

[0013]

[0014] Where: λ is the maximum and minimum threshold, N is the number of sampling points;

[0015] S22, reconstructing the denoised sampled signal to obtain a denoised signal; the reconstructed expression is as follows:

[0016] σ=mediab(s(z)) / 0.6745, z=0, 1,...,N-1;

[0017] Where: σ is the approximate variance of the noisy sampling signal, median is the median, s(z) is the data point of the component, and z is the current sampling point;

[0018] S23, calculating the energy value of the sampled signal; the calculation formula is as follows:

[0019]

[0020] Where: X 2 (z) is the sampling signal when the sampling point is z.

[0021] In step S3, the calculation formula for obtaining the maximum and minimum energy values ​​is as follows:

[0022]

[0023] Where: X i is the energy value of the i-th sampling signal, X max is the maximum energy of the sampled signal, X min is the minimum energy value of the sampled signal.

[0024] The optimal point set optimization initialization population in step S4 refers to initializing the starting position of individual whales by introducing the optimal point set method, as follows:

[0025] Let G s is a unit cube in S-dimensional Euclidean space, if a good point g∈G s , then the good point set is as follows:

[0026] P n (k)={({g1·k}, {g2·k},…,{g s ·k}), 1≤k≤n};

[0027] Where: P n (k) is the set of good points, {g s ·k} is to take g s k is the decimal part of the result, g is the optimal point, n is the number of samples, and k is the kth sample among n samples;

[0028] The deviation satisfies the following:

[0029]

[0030] in: is the deviation, C(g, ε) is a constant, ε is any positive number, and g is the optimal point;

[0031]

[0032] Where: g j is the best point of the jth dimension; j is the spatial dimension, and p is the smallest prime number;

[0033] For an unknown uniformly distributed object, the deviation of taking n points by randomly selecting points is as follows:

[0034] O(n -1 / 2 (loglogn) 1 / 2 );

[0035] The individual positions of the whale population based on the good point set method are mapped to the search space as follows:

[0036]

[0037] in: is the position of the kth whale in the jth dimensional space, ub j lb j are the upper and lower limits of the search range in the j-th dimensional space respectively.

[0038] The elite reverse learning strategy in step S4 refers to: by introducing the elite strategy, the initialized whale population is sorted by fitness, and individuals with higher fitness are selected as elite individuals, the reverse solution of the elite solution is calculated, and the individuals with better performance in the elite solution and the reverse solution are selected as elite individuals of the next generation population, as follows:

[0039] Assuming that the whales after each round of iteration are elite individuals, the expression is as follows:

[0040]

[0041] in: is the position vector of the i-th whale in the j-th dimensional space, D is the spatial dimension, N is the population size, is the position of the Nth whale in D-dimensional space;

[0042] The strategy formula for calculating the reverse solution of the elite solution is as follows:

[0043]

[0044]

[0045] in: and are all reverse solutions, c is a random number in the interval [0,1], ub j lb j are the upper and lower limits of the dynamic boundary respectively;

[0046] If the reverse solution exceeds the boundary, the reverse solution is reset by random generation; the reset formula is as follows:

[0047]

[0048] Among them: rand() is a random function.

[0049] The nonlinear parameter control in step S4 refers to: changing the variation of the convergence factor through the nonlinear convergence factor, strengthening the local development capability to jump out of the local optimal value, as follows:

[0050] The nonlinear convergence factor expression is as follows:

[0051]

[0052] Among them: a is the nonlinear convergence factor, b and e are control factors, T max is the maximum number of iterations, and t is the current number of iterations.

[0053] The random difference mutation strategy in step S4 refers to: using the random difference mutation strategy to perform random difference mutation disturbance on the whale population in the search phase, and obtaining the optimal position before and after the change again, as follows:

[0054] For the search phase, the random difference mutation strategy is as follows:

[0055] X(t+1)=δ1(X * (t)-X(t))+δ2(X rand (txt));

[0056] Where: X(t+1) is the whale position after each iteration, X * (t) is the optimal position, X(t) is the current position of the whale, δ1 and δ2 are both random numbers in the interval [0, 1], and X rand (t) is a randomly selected individual whale in the population.

[0057] The step S5 specifically includes:

[0058] S51, setting includes population size N and maximum number of iterations T max The initial variables are used to initialize the whale population using the best point set method and calculate the fitness;

[0059] S52, using the elite reverse learning strategy to select elite whale individuals, calculate the dynamic boundary of the elite whale individuals, and calculate the reverse solution to obtain the reverse population;

[0060] S53, compare the elite whale individuals and the reverse population, select the top N whale populations with better fitness as the new whale population to participate in the next iteration;

[0061] S54, updating the coefficient vector of the whale population, controlling the movement of the whale population, and updating the convergence factor; the expression of the coefficient vector is as follows:

[0062]

[0063] in: is the distance vector between the individual whale and its prey, is the position vector of the optimal solution currently obtained, is the position vector of the individual whale at the tth iteration, is the position vector of the individual whale at the t+1th iteration, are coefficient vectors;

[0064] The expression of the update convergence factor is as follows:

[0065]

[0066] in: is the reverse solution, c is a random number in the interval [0, 1], ub j lb j are the upper and lower limits of the dynamic boundary respectively;

[0067] S55, generating a random number p in the interval [0, 1], and determining the fishing method of the whale population; the determination result includes any one of the following:

[0068] The first type: when p < 0.5, the IWOA algorithm performs the search bracketing phase;

[0069] Among them, if When , a whale individual is randomly selected from the whale population for optimization, and the position of the whale individual is updated by the following formula:

[0070]

[0071] Where: X(t+1) is the individual position of the whale, D is the spatial dimension, X * (t) is the optimal position, e is the control factor;

[0072] like When , a whale individual is randomly selected from the whale population for optimization, and the position of the whale individual is updated by the following formula:

[0073]

[0074] Where: δ(||Yc i ||) is the Gaussian function, Y is the input vector, c i is the neuron center point of the hidden layer in the RBF neural network, ||Yc i || is the input vector Y to the center point c i The Euclidean distance, σ i is the Gaussian kernel width of the hidden layer in the RBF neural network;

[0075] Second: When p ≥ 0.5, the IWOA algorithm updates the position in a spiral manner according to the following formula:

[0076]

[0077] in: is a random vector on [0, 1], a is the convergence factor, and t is the number of iterations;

[0078] S56, iterating the IWOA algorithm and determining whether a maximum number of iterations has been reached; the determination result includes any one of the following:

[0079] The first type: If the number of iterations t<T max, then the whale population is randomly differentially perturbed according to the following formula, and after the number of iterations t = t + 1, it returns to step S4 to continue iterating;

[0080]

[0081] in: is the reverse solution, rand() is a random function;

[0082] The second type: If the number of iterations t<T max If it does not hold, then the global optimal solution is output.

[0083] A transmission line equipment fault classification system based on DWT-RBF, the system is applied to the above method, the system comprises:

[0084] Simulation system building module, used to build a DWT-RBF simulation system for three-phase series compensation power;

[0085] The signal decomposition module is used to obtain the sampling signal of the transient current when the transmission line fails, decompose the sampling signal using DWT, and then denoise it based on the improved wavelet threshold algorithm to obtain the energy value of the sampling signal;

[0086] A normalization module is used to normalize the energy value of the sampled signal and obtain the maximum and minimum values ​​of the energy value;

[0087] An algorithm optimization module is used to optimize the DWT-RBF simulation system based on the IWOA algorithm; the IWOA algorithm optimization includes: any one or any combination of the optimal point set optimization initialization population, elite reverse learning strategy, nonlinear parameter control, and random difference mutation strategy;

[0088] The fault classification module is used to input the energy value into the optimized DWT-RBF neural network based on the optimized DWT-RBF simulation system to complete the fault classification.

[0089] A transmission line equipment fault classification device based on DWT-RBF, the device comprising a processor and a memory;

[0090] The memory is used to store computer program code and transmit the computer program code to the processor;

[0091] The processor is used to execute the above-mentioned DWT-RBF-based power transmission line equipment fault classification method according to the instructions in the computer program code.

[0092] Compared with the prior art, the present invention has the following beneficial effects:

[0093] In a transmission line equipment fault classification method, system and device based on DWT-RBF of the present invention, the method first builds a simulation system of three-phase series compensation power and defines the fault type, then obtains the sampling signal of transient current when the transmission line fails, decomposes it using DWT, then denoises it based on an improved wavelet threshold algorithm, and obtains the energy value of the sampling signal, then performs normalization processing on it, and finally optimizes the DWT-RBF simulation system based on an IWOA algorithm, and inputs the energy value to complete the fault classification; in the application of this design, the sampling signal is decomposed by DWT and the fault characteristics are extracted, which speeds up the response speed of fault classification, and the wavelet threshold denoising algorithm is used for noise filtering to improve the recognition accuracy, and at the same time, the RBF model is optimized by the IWOA algorithm to improve the convergence speed and generalization fitting ability of the model, so that the present invention has better network performance and can accurately and quickly classify different types of transmission line equipment faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0094] Figure 1 It is a flow chart of the method steps of the present invention.

[0095] Figure 2 It is a structural schematic diagram of the three-phase series compensation power system in Example 1 of the present invention.

[0096] Figure 3 It is a schematic diagram of the wavelet three-level decomposition constructed by MRA in Example 1 of the present invention.

[0097] Figure 4 It is a schematic diagram of the fault classification processing structure in Example 1 of the present invention.

[0098] Figure 5 It is a schematic diagram of the fast Fourier transform of the A-phase fault signal in Example 1 of the present invention.

[0099] Figure 6 It is a schematic diagram of the wavelet three-level decomposition technology db4 of the A-phase fault signal in Example 1 of the present invention.

[0100] Figure 7 It is a schematic diagram of the relationship between the fault starting angle and the wavelet transform energy in Example 1 of the present invention.

[0101] Figure 8 It is a schematic diagram of the radial basis network structure in Example 1 of the present invention.

[0102] Fig. 9 Schematic diagram of network performance of training samples in phase A in Example 1 of the present invention.

[0103] Fig.10 It is a schematic diagram of the system structure of the present invention.

[0104] Fig.11It is a schematic diagram of the device structure of the present invention.

[0105] In the figure: simulation system construction module 1, signal decomposition module 2, normalization module 3, algorithm optimization module 4, fault classification module 5, processor 6, memory 7, computer program code 71. DETAILED DESCRIPTION

[0106] The present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0107] Embodiment 1:

[0108] See also Figure 1 , a transmission line equipment fault classification method based on DWT-RBF, including:

[0109] S1. Build a DWT-RBF simulation system for three-phase series compensation power;

[0110] See also Figure 2 The DWT-RBF simulation system of three-phase series compensation power is built in MATLAB or SIMULINK. It is a three-phase, 60Hz, 735kV power system that transmits power from a power plant consisting of 6 350MVA generators to an equivalent system through a 600km transmission line. The transmission line is divided into two 300km lines connected between busbars B1, B2 and B3; in order to increase the transmission capacity, each line is series compensated by a capacitor representing 40% of the line reactance. The two lines are also parallel compensated by a 330Mvar parallel reactance. The parallel and series compensation equipment is located in the B2 substation. The 300MVA-735 / 230kV transformer supplies power to the 230kV-250MW load. Each series compensation group is protected by a metal oxide varistor (MOV1 and MOV2). CB1 and CB2 represent two circuit breakers of line 1. The simulation time is 50μs, which is only a quarter cycle, and the fault resistance is equal to 1Ω.

[0111] In this embodiment, it is also necessary to define the fault types. Ten types of faults (ag, bg, cg, abg, ab, bcg, bc, acg, ca, abc) and four fault starting angles (0, 45, 90, 135) occur on the transmission line between the busbars (B1-B2) with a total length of 300 kilometers. All fault types and four fault locations are processed from the transmitting end distance (25 kilometers, 100 kilometers, 175 kilometers, 250 kilometers) before and after series compensation. Therefore, the total number of cases in this embodiment is 160 (10*4*4=160).

[0112] In this embodiment, the sampling signal of the transient current when the transmission line fails is the current signal and the fault time collected in real time by a current sensor installed on the transmission line or an online monitoring system.

[0113] S2, obtaining a sampling signal of transient current when a transmission line fails, decomposing the sampling signal using DWT, then denoising based on an improved wavelet threshold algorithm, and obtaining an energy value of the sampling signal;

[0114] In this embodiment, the principle of DWT signal decomposition is as follows:

[0115] DWT consists of a scaling function and a wavelet function, and the expressions are as follows:

[0116] φ jk =2 j / 2 φ 00 (2 j zk);

[0117]

[0118] Where: φ jk and are the scale function and wavelet function when the scale parameter is j and the translation parameter is k, respectively, z is the current sampling point, is the basis function of the scaling function, φ 00 is the basis function of the wavelet function, also called the mother wavelet;

[0119] Multiresolution analysis (MRA) provides different time and frequency resolutions at each analysis level. MRA uses a scaling function φ and a wavelet function So that the sampled signal can be expanded to:

[0120]

[0121] Where: X(z) is the sampling signal when the sampling point is z, a jk d jk They are the approximation coefficient and detail coefficient when the scale parameter is j and the translation parameter is k, collectively referred to as DWT coefficients;

[0122] The basic idea of ​​DWT is to decompose the signal into different scales, and perform low-pass and high-pass filtering on the sub-signals of various scales. The output of the high-pass filter contains the detail coefficient d of the input signal, and the output of the low-pass filter is the approximate coefficient a. The frequency resolution is improved by downsampling the signal of the filter, which is specifically expressed as follows:

[0123] The low-frequency component a at the jth scale when the sampling point is z j (z) and high frequency component d j (z), which is expressed as follows:

[0124]

[0125] Among them: a j+1 (z) is the low-frequency component at the j+1th scale when the sampling point is z, and G0 and H0 are coefficients, corresponding to the filter coefficients of the low-pass filter and the high-pass filter, respectively.

[0126] See also Figure 3 , the three-level decomposition of wavelets constructed by MRA is carried out by using db4 as the mother wavelet. The decomposition of the signal reaches the third level. In each analysis level, the signal is decomposed by applying continuous low-pass filter G0 and high-pass filter H0. Half of the samples can be eliminated after each filtering, and decomposition is achieved at the end line. Finally, detail coefficients d1, d2, d3 and approximate coefficient a3 can be obtained, thereby extracting time information and frequency information at the same time. Therefore, lower frequencies can achieve higher time resolution. Zero-phase wavelets like db4 have the smallest energy envelope width, and their energy is limited to the smallest time. Therefore, zero-phase wavelets have the largest resolution; the total energy is both the potential energy and the kinetic energy of the wavelet signal. Assuming that the wavelet signal is the output of a velocity detector, partial energy represents only kinetic energy, and total energy better represents the wavelet energy propagated through the earth; although the energy distribution is the same, the partial energy used (only the real signal) will produce a bias that makes the zero-phase wavelet better than other similar wavelets; in training and testing, instead of using detail information and approximate information, the total energy of each decomposition level is used as a feature for data classification, and the ED representing the detail energy of decomposition level q is used q and EA, which represents the approximate energy of the decomposition level p p The energy of wavelet coefficients varies at different scales depending on the input signal, including the approximate part and the detail part.

[0127] The transient current amplitude during line fault is weak and easily affected by noise. To ensure the accuracy of the fault identification algorithm, an improved wavelet threshold algorithm is used for noise filtering. The transient current signal collected by the equipment is composed of the superposition of the original signal and the noise signal. After wavelet transform, the approximate component and the detail component are obtained. Each component is still composed of the original signal and the noise signal, and the noise is mostly high-frequency components, located in the detail component of the wavelet transform. The modulus value of the original signal accounts for a larger proportion than the noise signal, and the actual noise is composed of multiple sources and can be regarded as the sum of variables with different independent probability distributions. According to the central limit theorem, the noise distribution will approach the Gaussian distribution. Therefore, the denoising process in this embodiment is as follows:

[0128] S21. Based on the normal distribution principle of Gaussian distribution, the sampling signal noise is distributed in the interval of [-3σ, 3σ]. The maximum and minimum thresholds are used to denoise the data of each layer component in the interval of [-3σ, 3σ]. The denoising expression is as follows:

[0129]

[0130] Where: λ is the maximum and minimum threshold, N is the number of sampling points;

[0131] S22, reconstructing the denoised sampled signal to obtain a denoised signal; the reconstructed expression is as follows:

[0132] σ=median(s(z)) / 0.6745, z=0, 1,...,N-1;

[0133] Where: σ is the approximate variance of the noisy sampling signal, median is the median of the data, s(z) is the data point of a certain layer component, and z is the current sampling point;

[0134] In the application of wavelet threshold denoising algorithm, it is necessary to consider not only the distribution range of noise, but also the influence of variance σ on the threshold. Here, the σ of each layer component is multiplied by the maximum and minimum thresholds of each layer to obtain a new threshold. The threshold calculation method can be divided into hard threshold method and soft threshold method. The hard threshold method calculates the threshold λ according to the calculation, sets the points with modulus values ​​less than λ in the wavelet component to be denoised to zero, and retains the points with modulus values ​​greater than λ; the soft threshold denoising method subtracts λ from the points with modulus values ​​greater than λ in the component, and sets the points with modulus values ​​less than λ to zero. Compared with the hard threshold method, the soft threshold method can remove noise more smoothly and ensure the restoration of the waveform. Therefore, the soft threshold method is preferably used for denoising in this technical solution.

[0135] S23, calculating the energy value of the sampled signal; the calculation formula is as follows:

[0136]

[0137] Where: X 2 (z) is the sampling signal when the sampling point is z.

[0138] S3, normalizing the energy value after decomposition and denoising to obtain the maximum and minimum values ​​of the energy value;

[0139] Furthermore, the calculation formula for obtaining the maximum and minimum energy values ​​is as follows:

[0140]

[0141] Where: X i is the energy value of the i-th sampling signal, X max is the maximum energy of the sampled signal, Xmin is the minimum energy value of the sampled signal.

[0142] See also Figure 4 After the energy value is normalized to between 0 and 1, it is input into the RBF network optimized by IWOA for fault classification. The output target of the detection is 0 when normal and 1 when faulty. For example, the output target of BCG fault is (0 11 1).

[0143] S4. Optimizing the DWT-RBF simulation system based on the IWOA algorithm; the IWOA algorithm optimization includes: any one or any combination of the optimal point set optimization initialization population, elite reverse learning strategy, nonlinear parameter control, and random difference mutation strategy;

[0144] In this embodiment, the RBF neural network principle is as follows:

[0145] RBF neural network is a typical multi-layer forward neural network. It uses a linear method to map problems that cannot be solved in low-dimensional space to high-dimensional space for solution. RBF neural network includes input layer, hidden layer and output layer, where the hidden layer is composed of RBF network neurons; RBF neural network is represented as follows:

[0146]

[0147] Where: Y is the input vector, Y∈R n ,ω i is the weight between the i-th hidden layer neuron and the output layer in the RBF neural network, c i is the center point of the i-th neuron in the RBF neural network, m is the number of neurons in the hidden layer, ||Yc i || is the input vector Y to the center point c i The Euclidean distance of i ||) is R + →The linear function of R is usually called the Gaussian function, which is expressed as follows:

[0148]

[0149] Where: i is the Gaussian kernel width of the hidden layer in the RBF neural network;

[0150] The training speed and accuracy of the RBF neural network prediction model are related to the linear weight ω of the neurons in the output layer i , the neuron center point c of the hidden layer i and the Gaussian kernel width σ of the hidden layer i Usually, clustering algorithms are used to first obtain c i and σ iThen, the least square method is used to obtain ω i However, this method will slow down the convergence of the neural network when obtaining the parameters of the RBF neural network, while IWOA has a strong optimization capability. Therefore, the use of IWOA to optimize these three parameters in the present invention can speed up the training speed of the prediction model and improve the prediction accuracy.

[0151] In this embodiment, the principle of the WOA algorithm is as follows:

[0152] WOA is a swarm intelligence optimization algorithm that simulates the hunting behavior of humpback whales. The basic model of the algorithm is described as follows: Assuming that there are only n whale individuals in the d-dimensional search space, the i-th individual at iteration t is expressed as Where T max is the maximum number of iterations.

[0153] The algorithm consists of three stages: the humpback whales surround the prey stage, the local search stage and the global search stage.

[0154] The humpback whale encirclement phase is shown below:

[0155]

[0156]

[0157] in: is the distance vector between the individual whale and its prey; is the position vector of the optimal solution currently obtained; is the position vector of the individual whale at the tth iteration; is the position vector of the individual whale at the t+1th iteration; are coefficient vectors used to control the swimming pattern of the whale; |·| is the absolute value.

[0158] In the above formula and The calculation process is as follows:

[0159]

[0160] in: is a random vector in the interval [0, 1], is the convergence factor, which decreases linearly from 2 to 0 with the increase of iteration number during the iteration process.

[0161] The local search phase is represented as follows:

[0162]

[0163] in: is the distance from the ith whale to its prey (the optimal solution currently obtained); b is a constant that defines the shape of the logarithmic spiral, and l is a random number in [-1, 1].

[0164] When a humpback whale approaches its prey, its shrinking and surrounding and spiral position updating behaviors are performed simultaneously. In this embodiment, it is assumed that the humpback whale has a 50% probability of shrinking and surrounding and spiral position updating, and its mathematical model description is as follows:

[0165]

[0166] Where: X(t+1) is the individual position of the whale, D is the spatial dimension, X * (t) is the optimal position and e is the control factor.

[0167] The global search phase is represented as follows:

[0168] The global search mechanism randomly selects individual whales for learning, updates each position, and combines Perform a global search; when When the whale performs random search and global search to avoid falling into the local optimum, the mathematical model at this stage is as follows:

[0169]

[0170] in: is the distance vector from a randomly selected whale individual to its prey, is the position vector of a randomly selected individual whale.

[0171] Although WOA has the advantages of fewer parameters, simple structure, and better convergence effect, it still has the following problems:

[0172] 1. Uneven distribution of population initialization: WOA randomly initializes the population, which may cause uneven distribution or excessive concentration, which is not conducive to finding the global optimal solution and affects the optimization efficiency and accuracy;

[0173] 2. Slow convergence speed: The parameter a changes linearly from 2 to 0, and the convergence speed of the algorithm slows down in the later stage;

[0174] 3. It is difficult to jump out of the local optimal solution: In the later stage of the algorithm, as the parameter A continues to decrease, the update range of the current optimal solution is also shrinking. Once it falls into the local optimal solution, it will be difficult to jump out.

[0175] In view of the above shortcomings, in this embodiment, the WOA algorithm is optimized by any one or any combination of the good point set optimization initialization population, elite reverse learning strategy, nonlinear parameter control, and random difference mutation strategy.

[0176] The optimal point set optimization initializes the population as follows:

[0177] In the swarm intelligence optimization algorithm, a good population initialization will optimize the algorithm's search for the global optimum and improve the accuracy of the optimal solution. The uniform distribution and diversity of the population are conducive to the improvement of the algorithm's optimization performance. Therefore, the introduction of the good point set is used to initialize the starting position of the individual whales, as follows:

[0178] Let G s is a unit cube in s-dimensional Euclidean space, if a good point g∈G s , then the good point set is as follows:

[0179] Pn(k)={({g1·k}, {g2·k},…,{g s ·k}), 1≤k≤n};

[0180] Where: P n (k) is the set of good points, {g s ·k} is to take g s k is the decimal part of the result, g is the optimal point, n is the number of samples, and k is the kth sample among n samples;

[0181] The deviation satisfies the following:

[0182]

[0183] in: is the deviation, C(g, ε) is a constant related to g and ε, ε is an arbitrary positive number, and g is the optimal point;

[0184]

[0185] Where: g j is the best point of the jth dimension; j is the spatial dimension, p is the smallest prime number that satisfies (p-3) / 2≥s;

[0186] For an unknown uniformly distributed object, the deviation of taking n points by randomly selecting points is as follows:

[0187] O(n -1 / 2 (loglogn) 1 / 2 );

[0188] From the formula of the randomly selected whale individual position vector in the global search mechanism, we can know that the deviation of selecting n points using the best point set is O(n -1+ε ). Therefore, the deviation of point selection using the good point set method is very small, only at the square root level of random point selection. Initializing the individual positions of whales through the good point set can increase the diversity of the population, improve the global search ability of the whale optimization algorithm to a certain extent, and accelerate the convergence speed of the algorithm.

[0189] The individual positions of the whale population based on the good point set method are mapped to the search space as follows:

[0190]

[0191] in: is the position of the kth whale in the jth dimensional space, ub j lb j are the upper and lower limits of the search range in the j-th dimensional space respectively.

[0192] The elite reverse learning strategy is as follows:

[0193] The basic idea of ​​reverse learning is to find the corresponding reverse solution based on the current solution, compare the fitness values ​​of the original solution and the reverse solution, and save the better solution; the introduction of reverse solution can expand the search space of the algorithm, but because reverse learning has a certain blindness, too much calculation of reverse solution will waste algorithm search time; therefore, the elite strategy is introduced to sort the initialized whale population by fitness, select individuals with higher fitness as elite individuals, and calculate their reverse solutions, and select the elite solutions and the better individuals in their reverse solutions as the elite individuals of the next generation of population, while ensuring that the population continues to approach the optimal value, and reduces unnecessary calculations; the details are as follows:

[0194] Assuming that the whales after each round of iteration are elite individuals, the expression is as follows:

[0195]

[0196] in: is the position vector of the i-th whale in the j-th dimensional space, D is the spatial dimension, N is the population size, is the position of the Nth whale in D-dimensional space;

[0197] The strategy formula for calculating the reverse solution of the elite solution is as follows:

[0198]

[0199] in: and are all reverse solutions, c is a random number in the interval [0,1], ub j lb j are the upper and lower limits of the dynamic boundary, The advantage of dynamic boundaries is that they can make up for the defect that fixed boundaries cannot save search experience, which helps to shorten the search time.

[0200] If the reverse solution exceeds the boundary, it is reset using random generation; the reset formula is as follows:

[0201]

[0202] Among them: rand() is a random function.

[0203] The nonlinear parameter control is as follows:

[0204] In the basic whale optimization algorithm, the global exploration and local development capabilities of the whale group mainly depend on the convergence factor a; in the early stage of the WOA algorithm, there are fewer optimal individuals in the whale group, and the global exploration capability of the population needs to be strengthened. In the later stage of the algorithm, the population is more likely to have local optimal problems, and it is necessary to change the amount of convergence factor change and strengthen the local development capability to jump out of the local optimal value; but since the convergence factor a decreases linearly from 2 to 0, it is impossible to balance the global exploration capability and local development capability of the algorithm well. Therefore, through the nonlinear convergence factor, the amount of convergence factor change is changed, and the local development capability is strengthened to jump out of the local optimal value, as follows:

[0205] The nonlinear convergence factor expression is as follows:

[0206]

[0207] Among them: a is the nonlinear convergence factor, b and e are control factors, T max is the maximum number of iterations, and t is the current number of iterations.

[0208] In the early stage of the algorithm, the convergence factor decreases rapidly, which enhances the global search capability of the population, while the value converges slowly in the later stage of iteration, which is more conducive to local development and improves the convergence accuracy of the algorithm.

[0209] The random difference mutation strategy is as follows:

[0210] Traditional WOA is prone to premature convergence. Therefore, the purpose of using the random differential mutation strategy is to prevent the whale algorithm from falling into the local optimum in the later development. The specific approach is: each individual whale obtains the current optimal position after the position is updated after encircling prey or preying with a bubble net. Before the number of iterations reaches the maximum value, the random differential mutation strategy is used to perform random differential mutation perturbations on the population to obtain the optimal position before and after the change again, thereby overcoming the disadvantage of traditional WOA falling into the local optimum and accelerating the convergence of the algorithm. For the search stage, the random differential mutation strategy is shown as follows:

[0211] X(t+1)=δ1(X * (t)-X(t))+δ2(X rand (txt));

[0212] Where: X(t+1) is the whale position after each iteration, X *(t) is the optimal position, X(t) is the current position of the whale, δ1 and δ2 are both random numbers in the interval [0, 1], and X rand (t) is a randomly selected individual whale in the population.

[0213] S5. Based on the optimized DWT-RBF simulation system, the energy value is input into the optimized DWT-RBF neural network to complete the fault classification.

[0214] Furthermore, in this embodiment, the IWOA algorithm process is as follows:

[0215] S51, setting includes population size N and maximum number of iterations T max The initial variables are used to initialize the whale population using the best point set method and calculate the fitness;

[0216] S52, using the elite reverse learning strategy to select elite whale individuals, calculate the dynamic boundary of the elite whale individuals, and calculate the reverse solution to obtain the reverse population;

[0217] S53, compare the elite whale individuals and the reverse population, select the top N whale populations with better fitness as the new whale population to participate in the next iteration;

[0218] S54, updating the coefficient vector of the whale population, controlling the movement of the whale population, and updating the convergence factor; the expression of the coefficient vector is as follows:

[0219]

[0220] in: is the distance vector between the individual whale and its prey, is the position vector of the optimal solution currently obtained, is the position vector of the individual whale at the tth iteration, is the position vector of the individual whale at the t+1th iteration, are coefficient vectors;

[0221] The expression of the update convergence factor is as follows:

[0222]

[0223] in: is the reverse solution, c is a random number in the interval [0, 1], ub j lb j are the upper and lower limits of the dynamic boundary respectively;

[0224] S55, generating a random number p in the interval [0,1], and determining the whale fishing method; the determination result includes any one of the following:

[0225] The first one: when p<0.5, the IWOA algorithm performs the search bracketing phase;

[0226] Among them, if When , a whale individual is randomly selected from the whale population for optimization, and the position of the whale individual is updated by the following formula:

[0227]

[0228] Where: X(t+1) is the individual position of the whale, D is the spatial dimension, X * (t) is the optimal position, e is the control factor;

[0229] like When , a whale individual is randomly selected from the whale population for optimization, and the position of the whale individual is updated by the following formula:

[0230]

[0231] Where: δ(||Yc i ||) is the Gaussian function, Y is the input vector, c i is the neuron center point of the hidden layer in the RBF neural network, ||Yc i || is the input vector Y to the center point c i The Euclidean distance, σ i is the Gaussian kernel width of the hidden layer in the RBF neural network;

[0232] Second: When p ≥ 0.5, the IWOA algorithm updates the position in a spiral manner according to the following formula:

[0233]

[0234] in: is a random vector on [0, 1], a is the convergence factor, and t is the number of iterations;

[0235] S56, iterating the IWOA algorithm and determining whether a maximum number of iterations has been reached; the determination result includes any one of the following:

[0236] The first type: If the number of iterations t<T max , then the whale population is randomly differentially perturbed according to the following formula, and after the number of iterations t = t + 1, it returns to step S4 to continue iterating;

[0237]

[0238] in: is the reverse solution, rand() is a random function;

[0239] The second type: If the number of iterations is t <Tmax If it does not hold, then the global optimal solution is output.

[0240] In this embodiment, the effectiveness of the method is verified by simulating the current signal when the AG single-line grounding fault occurs.

[0241] The Fast Fourier Transform (FFT) has limited ability to analyze signals and give time-frequency information because it uses the global function ejwt to convert the signal from the time domain to the frequency domain. Therefore, all information in the time domain is lost, which can help detect the start time and duration of the interference. The Fast Fourier Transform of the A phase fault signal is as follows: Figure 5 As shown (THD stands for total harmonic distortion), the wavelet three-level decomposition technology db4 of the signal is implemented as Figure 6 As shown, it can be used to locate the distortion event in time, and even if the oscillation phenomenon decays to a very small value, the exact end time of the distortion event can be detected.

[0242] The energy of the current signal under normal fault-free conditions is shown in the following table:

[0243] Phase A B C normal 1.26E+05 1.99E+05 1.63E+05

[0244] From the above table, it can be seen that the energy values ​​of different phases are very close or almost the same, and these values ​​are used as a reference for different fault effects of noise at signal energy values; in the case of a single-phase ground fault, although the other two phases have slight changes, the energy of the affected line increases sharply from its normal state.

[0245] The signal energy in case of single-phase grounding fault is shown in the following table:

[0246]

[0247] From the table above, we can see that, for example, when an AG fault occurs, the maximum energy appears in phase A; the fault start angle affects the current energy, but for the same fault type and the same location, such as when an AG single-line grounding fault occurs at a distance of 25 km, the fault start angle has a greater impact on the energy value of the three-phase current (A) signal. The relationship between the fault start angle and the wavelet transform energy is shown in the figure below: Figure 7 As shown in the figure, when there is a ground fault in the double-circuit line, although the normal phase changes slightly, the energy of the affected double-circuit line increases sharply compared to the normal state.

[0248] The signal energy of a double-circuit line w / wo grounding fault is shown in the following table:

[0249]

[0250] It can be seen from the above table that, for example, when ABG or AB fails, the minimum response to the fault increases to the normal phase C.

[0251] The signal energy when three-wire fault occurs is shown in the following table:

[0252]

[0253] Furthermore, for the exact design method, the number of hidden neurons is equal to the number of training samples n h =N=144, the error of the training sample is zero, but for a larger neural network system, the number of hidden neurons will be too large and unacceptable. However, the RBF network was established using IWOA. After testing, the distribution degree of the network was 0.065 and the allowable error was 0.01. The radial basis network structure established in this embodiment is shown in the figure Figure 8 As shown in Figure 2, the proposed network has three input signals, 108 hidden neurons, and 4 neurons in the output layer. The network performance of the training sample in phase A is shown in Figure 2. Fig. 9 As shown in the figure, it can be seen that the training objective of this stage is very close to the output result of the network.

[0254] In order to test the generalization ability of the RBF network, we selected other samples that did not exist in the training set for testing. The test results are shown in the following table:

[0255]

[0256] The cases in the above table show that all fault elements can be correctly identified, so the network performance of this technical solution is good.

[0257] The transmission line equipment generally includes line towers, conductors, insulators, line fittings, guy wires and tower foundations, grounding devices, lightning arresters, circuit breakers, transformers, voltage transformers, etc. In this embodiment, the equipment that is prone to short circuit failures, such as conductors, transformers, switches, insulators, and circuit breakers, is mainly targeted. After a short circuit, the signal characteristics will show a sudden increase in current, frequency oscillation, etc. The fault type of the transmission circuit equipment is identified by screening and separating the signal characteristics.

[0258] Embodiment 2:

[0259] A transmission line equipment fault classification system based on DWT-RBF, the system is applied to the method described in embodiment 1, the system comprising:

[0260] Simulation system building module 1 is used to build a DWT-RBF simulation system for three-phase series compensation power and define fault types;

[0261] The signal decomposition module 2 is used to obtain the sampling signal of the transient current when the transmission line fails, decompose the sampling signal using DWT, and then denoise it based on the improved wavelet threshold algorithm, and obtain the energy value of the sampling signal;

[0262] Furthermore, the signal decomposition module 2 performs denoising according to the relevant steps of steps S21-S23 in embodiment 1:

[0263] Normalization module 3, used for normalizing the energy value after decomposition and denoising to obtain the maximum and minimum values ​​of the energy value;

[0264] Further, the normalization module 3 performs calculation according to the relevant steps of step S3 in embodiment 1;

[0265] Algorithm optimization module 4, used for optimizing the DWT-RBF simulation system based on the IWOA algorithm; the IWOA algorithm optimization includes: any one or any combination of the optimal point set optimization initialization population, elite reverse learning strategy, nonlinear parameter control, and random difference mutation strategy;

[0266] The algorithm optimization module 4 performs optimization according to the relevant steps of step S4 in embodiment 1:

[0267] The fault classification module 5 is used to input the energy value into the optimized DWT-RBF neural network based on the optimized DWT-RBF simulation system to complete the fault classification.

[0268] Further, the fault classification module 5 performs fault classification according to steps S51-S56 in embodiment 1:

[0269] Embodiment 3:

[0270] A transmission line equipment fault classification device based on DWT-RBF, the device comprising a processor 6 and a memory 7;

[0271] The memory 7 is used to store computer program code 71 and transmit the computer program code 71 to the processor 6;

[0272] The processor 6 is used to execute the DWT-RBF-based power transmission line equipment fault classification method described in Example 1 according to the instructions in the computer program code 71.

[0273] This embodiment also includes a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed on a computer, the DWT-RBF-based transmission line fault classification method described in Example 1 is implemented.

[0274] Generally speaking, the computer instructions for implementing the method of the present invention may be carried in any combination of one or more computer-readable storage media. Non-transitory computer-readable storage media may include any computer-readable media, except for the signal itself that is temporarily propagating.

[0275] The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EKROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, device, or device.

[0276] Computer program code for performing the operation of the present invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, SMalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages, in particular, Python suitable for neural network computing and platform frameworks based on TensorFlow, PyTorch, etc. can be used. The program code can be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer or to an external computer (for example, using an Internet service provider to connect via the Internet) through any type of network, including a local area network (LAN) or a wide area network (WAN).

[0277] The above-mentioned device and non-transitory computer-readable storage medium can refer to the specific description of a DWT-RBF-based transmission line equipment fault classification method and its beneficial effects, which will not be repeated here.

[0278] Although the embodiments of the present invention have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A transmission line equipment fault classification method based on DWT-RBF, characterized in that: include: S1. Build a DWT-RBF simulation system for three-phase series compensation power; S2, obtaining a sampling signal of transient current when a transmission line fails, decomposing the sampling signal using DWT, then denoising based on an improved wavelet threshold algorithm, and obtaining an energy value of the sampling signal; S3, normalizing the energy value of the sampled signal, and obtaining the maximum and minimum values ​​of the energy value; S4, Optimize DWT-RBF simulation system based on IWOA algorithm; The IWOA algorithm optimization includes: any one or any combination of the optimal point set optimization initialization population, elite reverse learning strategy, nonlinear parameter control, and random difference mutation strategy; S5. Based on the optimized DWT-RBF simulation system, the energy value is input into the optimized DWT-RBF neural network to complete the fault classification.

2. The method for classifying power transmission line equipment faults based on DWT-RBF according to claim 1, characterized in that: In step S2, the improved wavelet threshold algorithm denoising specifically includes: S21. Based on the normal distribution principle of Gaussian distribution, the sampling signal noise is distributed in the interval of [-3σ, 3σ]. The maximum and minimum thresholds are used to denoise the sampling data of each layer component in the interval of [-3σ, 3σ]. The denoising expression is as follows: Where: λ is the maximum and minimum threshold, N is the number of sampling points; S22, reconstructing the denoised sampled signal to obtain a denoised signal; the reconstructed expression is as follows: σ=median(s(z)) / 0.6745, z=0, 1,...,N-1; Where: σ is the approximate variance of the noisy sampling signal, median is the median, s(z) is the data point of the component, and z is the current sampling point; S23, calculating the energy value of the sampled signal; the calculation formula is as follows: Where: X 2 (z) is the sampling signal when the sampling point is z.

3. The method for classifying power transmission line equipment faults based on DWT-RBF according to claim 1, characterized in that: In step S3, the calculation formula for obtaining the maximum and minimum energy values ​​is as follows: Where: X i is the energy value of the i-th sampling signal, X max is the maximum energy of the sampled signal, X min is the minimum energy value of the sampled signal.

4. The method for classifying power transmission line equipment faults based on DWT-RBF according to claim 1, characterized in that: The optimal point set optimization initialization population in step S4 refers to initializing the starting position of individual whales by introducing the optimal point set method, as follows: Let G s is a unit cube in s-dimensional Euclidean space, if a good point g∈G s , then the good point set is as follows: P n (k)={({g1·k},{g2·k},…,{g s ·k}),1≤k≤n); Where: P n (k) is the set of good points, {g s ·k} is to take g s k is the decimal part of the result, g is the optimal point, n is the number of samples, and k is the kth sample among n samples; The deviation satisfies the following: in: is the deviation, C(g, ε) is a constant, ε is any positive number, and g is the optimal point; Where: g j is the best point of the jth dimension; j is the spatial dimension, and p is the smallest prime number; For an unknown uniformly distributed object, the deviation of taking n points by randomly selecting points is as follows: O(n -1 / 2 (loglogn) 1 / 2 ); The individual positions of the whale population based on the good point set method are mapped to the search space as follows: in: is the position of the kth whale in the jth dimensional space, ub j lb j are the upper and lower limits of the search range in the j-th dimensional space respectively.

5. The method for classifying power transmission line equipment faults based on DWT-RBF according to claim 1, characterized in that: The elite reverse learning strategy in step S4 refers to: by introducing the elite strategy, the initialized whale population is sorted by fitness, and individuals with higher fitness are selected as elite individuals, the reverse solution of the elite solution is calculated, and the individuals with better performance in the elite solution and the reverse solution are selected as elite individuals of the next generation population, as follows: Assuming that the whales after each round of iteration are elite individuals, the expression is as follows: in: is the position vector of the i-th whale in the j-th dimensional space, D is the spatial dimension, N is the population size, is the position of the Nth whale in D-dimensional space; The strategy formula for calculating the reverse solution of the elite solution is as follows: in: and are all reverse solutions, c is a random number in the interval [0, 1], ub j lb j are the upper and lower limits of the dynamic boundary respectively; If the reverse solution exceeds the boundary, the reverse solution is reset by random generation; the reset formula is as follows: Among them: rand() is a random function.

6. The method for classifying power transmission line equipment faults based on DWT-RBF according to claim 1, characterized in that: The nonlinear parameter control in step S4 refers to: changing the variation of the convergence factor through the nonlinear convergence factor, strengthening the local development capability to jump out of the local optimal value, as follows: The nonlinear convergence factor expression is as follows: Among them: a is the nonlinear convergence factor, b and e are control factors, T max is the maximum number of iterations, and t is the current number of iterations.

7. The method for classifying power transmission line equipment faults based on DWT-RBF according to claim 1, characterized in that: The random difference mutation strategy in step S4 refers to: using the random difference mutation strategy to perform random difference mutation disturbance on the whale population in the search phase, and obtaining the optimal position before and after the change again, as follows: For the search phase, the random difference mutation strategy is as follows: X(t+1)=δ1(X * (t)-X(t))+δ2(X rand (t)-X(t)); Where: X(t+1) is the whale position after each iteration, X * (t) is the optimal position, X(t) is the current position of the whale, δ1 and δ2 are both random numbers in the interval [0, 1], and X rand (t) is a randomly selected individual whale in the population.

8. The method for classifying power transmission line equipment faults based on DWT-RBF according to claim 1, characterized in that: The step S5 specifically includes: S51, setting includes population size N and maximum number of iterations T max The initial variables are used to initialize the whale population using the best point set method and calculate the fitness; S52, using the elite reverse learning strategy to select elite whale individuals, calculate the dynamic boundary of the elite whale individuals, and calculate the reverse solution to obtain the reverse population; S53, compare the elite whale individuals and the reverse population, select the top N whale populations with better fitness as the new whale population to participate in the next iteration; S54, updating the coefficient vector of the whale population, controlling the movement of the whale population, and updating the convergence factor; the expression of the coefficient vector is as follows: in: is the distance vector between the individual whale and its prey, is the position vector of the optimal solution currently obtained, is the position vector of the individual whale at the tth iteration, is the position vector of the individual whale at the t+1th iteration, are coefficient vectors; The expression of the update convergence factor is as follows: in: is the reverse solution, c is a random number in the interval [0, 1], ub j lb j are the upper and lower limits of the dynamic boundary respectively; S55, generating a random number p in the interval [0, 1], and determining the fishing method of the whale population; the determination result includes any one of the following: The first one: when p<0.5, the IWOA algorithm performs the search bracketing phase; Among them, if When , a whale individual is randomly selected from the whale population for optimization, and the position of the whale individual is updated by the following formula: Where: X(t+1) is the individual position of the whale, D is the spatial dimension, X*(t) is the optimal position, and e is the control factor; like When , a whale individual is randomly selected from the whale population for optimization, and the position of the whale individual is updated by the following formula: Where: δ(||Yc i ||) is the Gaussian function, Y is the input vector, c i is the neuron center point of the hidden layer in the RBF neural network, ||Yc i || is the input vector Y to the center point c i The Euclidean distance, σ i is the Gaussian kernel width of the hidden layer in the RBF neural network; Second: When p ≥ 0.5, the IWOA algorithm updates the position in a spiral manner according to the following formula: in: is a random vector on [0, 1], a is the convergence factor, and t is the number of iterations; S56, iterating the IWOA algorithm and determining whether a maximum number of iterations has been reached; the determination result includes any one of the following: The first one: If the number of iterations is t <T max , then the whale population is randomly differentially perturbed according to the following formula, and after the number of iterations t = t + 1, it returns to step S4 to continue iterating; in: is the reverse solution, rand() is a random function; The second type: If the number of iterations is t <T max If it does not hold, then the global optimal solution is output.

9. A transmission line equipment fault classification system based on DWT-RBF, characterized by: The system is applied to the method described in any one of claims 1 to 8, and the system comprises: A simulation system building module (1) is used to build a DWT-RBF simulation system for three-phase series compensation power; A signal decomposition module (2) is used to obtain a sampling signal of a transient current when a transmission line fails, decompose the sampling signal using DWT, and then perform denoising based on an improved wavelet threshold algorithm to obtain an energy value of the sampling signal; A normalization module (3), used for normalizing the energy value of the sampled signal and obtaining the maximum and minimum values ​​of the energy value; An algorithm optimization module (4) is used to optimize the DWT-RBF simulation system based on an IWOA algorithm; the IWOA algorithm optimization includes: any one or any combination of good point set optimization initialization population, elite reverse learning strategy, nonlinear parameter control, and random difference mutation strategy; The fault classification module (5) is used to input the energy value into the optimized DWT-RBF neural network based on the optimized DWT-RBF simulation system to complete the fault classification.

10. A transmission line equipment fault classification device based on DWT-RBF, characterized by: The device comprises a processor (6) and a memory (7); The memory (7) is used to store computer program code (71) and transmit the computer program code (71) to the processor (6); The processor (6) is used to execute the DWT-RBF-based power transmission line equipment fault classification method according to any one of claims 1 to 8 according to the instructions in the computer program code (71).

Images