Fault Location Method and System for Distribution Network Based on IELM-VMD Algorithm

Abstract

The present invention belongs to the technical field of power transmission, and provides a distribution network fault location method and system based on the IELM-VMD algorithm. The method includes determining ranging nodes based on the fault section location result and the verification mechanism; obtaining the three-phase fault signals of the sampling points based on the selected ranging nodes; performing phase-mode transformation on the three-phase fault signals, obtaining the modal components of the three-phase fault signals through Clarke transformation, and selecting the modal components according to the fault type and selection rules; based on the selected modal components, performing modal decomposition on the signals using the IELM-VMD algorithm, taking the VMD algorithm as the kernel, using the envelope entropy of the IMF components as the evaluation function, and optimizing and solving the decomposition layer number and penalty factor of VMD through the double-layer structure of the IELM algorithm to obtain the IMF components; based on the IMF components, extracting the natural frequencies of the IMF components, and calculating the fault distance based on the relationship between the natural frequencies of the IMF components and the fault distance.

Description

Technical Field

[0001] The present invention belongs to the technical field of power transmission, and particularly relates to a distribution network fault location method and system based on the IELM-VMD algorithm. Background Art

[0002] The statements in this part only provide background technical information related to the present invention and do not necessarily constitute prior art.

[0003] When the information uploaded by the FTUs at both ends of the fault section is distorted, there is a possibility of misjudgment in the distribution network fault section location based on FTUs. After the fault section location is completed, further determining the fault location through fault ranging has the following three main advantages: (1) It can determine the exact fault location, facilitating maintenance personnel for maintenance and reducing the workload of front-line personnel; (2) It can verify the fault section location result, so that when the fault section location algorithm misjudges, the fault section can be corrected in time, thereby avoiding the expansion of the fault when restoring power supply to non-fault areas later; (3) In the actual distribution network, the parameters of each section of the line are not exactly the same. Since the wave velocity is affected by the line parameters, when traveling waves propagate on different lines, they will show different speeds. When the traveling wave sampling points and the fault point span multiple lines with different parameters, it will inevitably lead to a decrease in ranging accuracy. Therefore, first determining the approximate range of the fault through fault section location and then through fault ranging can effectively reduce the possibility of the fault traveling wave crossing lines with different parameters, thereby improving the ranging accuracy.

[0004] For the distribution network fault ranging method based on the natural frequency of traveling waves, the key lies in how to accurately extract the main frequency of the signal. The traveling wave signal is often composed of countless traveling waves with specific frequencies superimposed. If directly extracting the main frequency of the signal based on the fault signal, it is difficult to accurately extract the main frequency of the signal due to the spectral aliasing phenomenon of the fault signal. As an adaptive signal processing method, the VMD algorithm can decompose the original signal into several independent components. Compared with other signal processing methods, it has the advantages of weak mode mixing and few false components, and can accurately separate the components containing the main frequency of the signal. However, the VMD algorithm requires manual selection of the decomposition layer number and the penalty factor, which not only has a low automation level, but also spectral aliasing may be caused by improper selection. Summary of the Invention

[0005] To solve at least one of the technical problems existing in the above-mentioned background art, the present invention provides a distribution network fault location method and system based on the IELM-VMD algorithm, which uses the envelope entropy of IMF components as an evaluation function, and optimally solves the parameters of VMD through the IELM algorithm, thereby avoiding manual setting of the parameters of the VMD algorithm, reducing the spectral aliasing and over-decomposition phenomena caused by improper manual setting, and improving the automation level and accuracy of modal decomposition.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] The first aspect of the present invention provides a distribution network fault location method based on the IELM-VMD algorithm, including: determining a ranging node based on the fault section location result and the verification mechanism;

[0008] Based on the selected ranging node, acquiring three-phase fault signals of sampling points;

[0009] Performing phase-mode transformation on the three-phase fault signals, obtaining the modal components of the three-phase fault signals through Clarke transformation, and selecting the modal components according to the fault type and selection rules;

[0010] Based on the selected modal components, performing modal decomposition on the signals using the IELM-VMD algorithm, taking the VMD algorithm as the kernel, using the envelope entropy of IMF components as the evaluation function, and optimally solving the decomposition layer number and penalty factor of VMD through the double-layer structure of the IELM algorithm to obtain IMF components;

[0011] Based on the IMF components, extracting the natural frequencies of the IMF components, and calculating the fault distance based on the relationship between the natural frequencies of the IMF components and the fault distance.

[0012] The second aspect of the present invention provides a distribution network fault location system based on the IELM-VMD algorithm, including:

[0013] A ranging node selection module for determining a ranging node based on the fault section location result and the verification mechanism;

[0014] A fault signal acquisition module for acquiring three-phase fault signals of sampling points based on the selected ranging node;

[0015] A modal component acquisition module for performing phase-mode transformation on the three-phase fault signals, obtaining the modal components of the three-phase fault signals through Clarke transformation, and selecting the modal components according to the fault type and selection rules;

[0016] A modal decomposition module, which is used to perform modal decomposition on a signal based on selected modal components by using the IELM-VMD algorithm. Taking the VMD algorithm as the kernel and the envelope entropy of the IMF components as the evaluation function, the decomposition layer number and penalty factor of VMD are optimized and solved through the double-layer structure of the IELM algorithm to obtain the IMF components.

[0017] A fault location module, which is used to extract the natural frequencies of the IMF components based on the IMF components and calculate the fault distance based on the relationship between the natural frequencies of the IMF components and the fault distance.

[0018] The third aspect of the present invention provides a computer-readable storage medium.

[0019] A computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the steps in the above-mentioned distribution network fault location method based on the IELM-VMD algorithm.

[0020] The fourth aspect of the present invention provides a computer device.

[0021] A computer device includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the steps in the above-mentioned distribution network fault location method based on the IELM-VMD algorithm.

[0022] Compared with the prior art, the beneficial effects of the present invention are:

[0023] The present invention proposes the IELM-VMD algorithm, whose main idea is to use the envelope entropy of the IMF components as the evaluation function, and optimize and solve the parameters of VMD through the IELM algorithm, so as to avoid manually setting the parameters of the VMD algorithm, reduce the spectrum aliasing and over-decomposition phenomena caused by improper manual setting, improve the automation level and accuracy of modal decomposition, thus achieving the best decomposition effect on the signal, and then accurately extracting the natural frequency of the traveling wave and realizing the accurate calculation of the fault point location.

[0024] The present invention can realize the automatic optimization setting of parameters, and also has a good modal decomposition effect, effectively avoiding the over-decomposition and modal aliasing phenomena, and improving the automation level of the algorithm. Moreover, it has good ranging accuracy when facing different types of faults and transition resistances;

[0025] The proposed verification mechanism of the present invention can effectively complete the verification of the fault section location result. Although there is a positive error in fault location and there is still a certain range of verification dead zones, the dead zone range is within an acceptable level, and the section junction points can be checked by means of line patrol. Description of the Drawings

[0026] The accompanying drawings of the specification, which form a part of the present invention, are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention.

[0027] Figure 1 It is the distributed parameter model of the transmission line for the embodiment of the present invention;

[0028] Figure 2 It is the refraction and reflection of traveling waves for the embodiment of the present invention;

[0029] Figure 3 It is the multiple refraction and reflection of traveling waves for the embodiment of the present invention;

[0030] Figure 4 It is the fault location flowchart based on the IELM-VMD algorithm for the embodiment of the present invention;

[0031] Figure 5 It is the schematic diagram of the ranging node positions of the hybrid line under unidirectional power flow for the embodiment of the present invention;

[0032] Figure 6 It is the schematic diagram of the ranging node positions of the hybrid line under bidirectional power flow for the embodiment of the present invention;

[0033] Figure 7 It is the fault model of the double-ended power supply line for the embodiment of the present invention;

[0034] Figure 8 It is the equivalent model of the two-port network of the fault of the double-ended power supply line for the embodiment of the present invention;

[0035] Figure 9 It is the flowchart of modal decomposition based on the IELM-VMD algorithm for the embodiment of the present invention;

[0036] Figure 10 It is the schematic diagram of the modal decomposition effect of the EMD algorithm for the embodiment of the present invention, where A is the original signal and B is the high-frequency harmonic IMF component;

[0037] Figure 11 It is the schematic diagram of the modal decomposition effect of the VMD algorithm for the embodiment of the present invention; where A is the original signal and the fundamental frequency IMF component, and B is the high-frequency harmonic IMF component;

[0038] Figure 12 It is the schematic diagram of the modal decomposition effect of the IELM-VMD algorithm for the embodiment of the present invention, where A is the original signal and the fundamental frequency IMF component, and B is the high-frequency harmonic IMF component;

[0039] Figure 13 It is the power spectrum extracted by the MUSIC algorithm for the embodiment of the present invention;

[0040] Figure 14This is a partial enlarged view of the IEEE 33-node dual-power distribution network according to an embodiment of the present invention. Detailed implementation manners

[0041] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0042] It should be noted that the following detailed description is illustrative and is intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which the present invention belongs.

[0043] It should be noted that the terms used herein are only for describing specific implementation manners and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular forms are also intended to include the plural forms. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0044] Term explanation:

[0045] A traveling wave refers to a transmission state of an electromagnetic wave on a transmission line. As Figure 1 shown, the transmission line can be equivalently regarded as an overall composed of countless micro-elements. The unit capacitance near the power supply side will be charged first, and then discharged through the unit inductance to the next unit capacitance. Through the continuous charging and discharging of each unit capacitance, the voltage propagates along the transmission line in the form of a traveling wave; similarly, a current waveform is formed through the continuous charging and discharging of each unit inductance.

[0046] When a traveling wave encounters an impedance mismatch point during transmission, to ensure the conservation of electric and magnetic field energies per unit length on both sides of this point, there is a phenomenon of energy redistribution when the traveling wave crosses this point, and this phenomenon is called the refraction and reflection of the traveling wave.

[0047] Figure 2 In, if the wave impedance of line 1 is Z1 and the wave impedance of line 2 is Z2. Then point A is an impedance mismatch point. When the incident wave arrives at point A from line 1, refraction and reflection occur at point A. When the incident wave is a voltage traveling wave, the refraction coefficient α U and the reflection coefficient β U are respectively:

[0048]

[0049] When the incident wave is a current traveling wave, the refraction coefficient α I and the reflection coefficient β I are respectively:

[0050]

[0051] The folding and reflection of traveling waves do not terminate after only a single occurrence, but rather occur infinitely many times at impedance mismatch points. A schematic diagram of the multiple folding and reflection of the incident wave in the line is as Figure 3 shown.

[0052] After a fault occurs on the transmission line, the transient fault traveling wave will Figure 3 go through the process shown in [figure reference]. The fault traveling wave undergoes infinitely many foldings and reflections and continuous superposition, ultimately forming the fault voltage and current signals. Therefore, the fault voltage and current signals in the frequency domain are manifested as the sum of frequencies in the form of infinitely many harmonics, and these frequencies are also called natural frequencies. Due to the numerous impedance mismatch points in the line, the composition of the natural frequencies is obviously very complex, but one thing is certain. That is, if sampling is performed at a certain point, the component with the highest proportion in the natural frequencies must reflect the impedance mismatch point closest to that point, because the traveling wave has the fastest round-trip speed between these two points and must have the highest energy within a limited time. Therefore, it is feasible to estimate the fault distance through the natural frequencies.

[0053] The VMD algorithm is a new time-frequency analysis method proposed by Dragomiretskiy et al. from the University of California, Los Angeles in 2014. Similar to signal decomposition methods such as Empirical Mode Decomposition (EMD), it can decompose a multi-component composite signal into multiple single-component amplitude-modulated and frequency-modulated signals. Its basic idea is to assume that any signal is formed by the superposition of several Intrinsic Mode Functions (IMFs), and all IMF components are signals with specific central frequencies and narrow bandwidths. Based on this characteristic, the VMD algorithm extracts IMF components within a variational framework, which mainly includes two aspects: one is the construction of the variational problem, establishing a variational optimization problem in the frequency domain; the other is the solution of the variational problem, solving the variational optimization problem in the frequency domain to obtain the central frequencies and bandwidths of each IMF component. Each IMF component is constructed through the central frequency, bandwidth, and corresponding constraints, thereby realizing the frequency-domain analysis of the original signal and the effective separation of each component.

[0054] Embodiment 1

[0055] Combined with Figure 4 shown in [figure reference], this embodiment provides a distribution network fault location method based on the IELM-VMD algorithm, including the following steps:

[0056] Step 1: Determine the ranging node based on the fault section location result;

[0057] Specifically, according to the positioning result of the fault section in the distribution network, the connection nodes between lines with different parameters and the nodes near the power source side are used as ranging nodes.

[0058] In this embodiment, fault ranging in the distribution network is performed; the fault ranging result of the distribution network is compared with the actual length of the transmission line in the distribution network to verify the positioning result of the fault section in the distribution network.

[0059] In an actual distribution network, the types of transmission lines in each section may vary, and the line parameters will affect the propagation speed of traveling waves on the lines. To minimize the error caused by the change in wave speed when traveling waves cross lines with different parameters, in this embodiment, according to the positioning result of the fault section in the distribution network, the connection nodes between lines with different parameters and the nodes near the power source side are used as ranging nodes, and fault ranging is performed at these nodes. Based on the verification mechanism, the positioning result of the fault section in the distribution network is verified by comparing the ranging result with the actual length of the transmission line. Specifically, after a fault occurs in the distribution network, the fault section positioning algorithm will first judge the fault section according to the information uploaded by the FTU. However, when the information uploaded by the FTUs at both ends of the fault section is distorted, there is a possibility of misjudgment in the fault section positioning based on the FTU. The present invention selects ranging nodes based on the fault section positioning result, and at these nodes, the distance from the ranging node to the fault point is calculated by the algorithm proposed in the present invention. The obtained ranging result is compared with the actual length parameter of the distribution network transmission line to check whether it conforms to the fault section positioning result, so as to verify the fault section positioning result. Figure 5 And Figure 6 take... as an example, the verification mechanism in two cases is explained respectively.

[0060] (1) Unidirectional power flow system

[0061] As Figure 5 shown, this system is a unidirectional power flow system, which contains two types of transmission lines. N1 and N3 are ranging nodes. From the F4-4 example, it can be seen that when the information uploaded by the FTUs at both ends of the fault section is distorted, it may lead to an incorrect judgment of the fault section, and the incorrect section must be on both sides of the correct section. Assume Figure 5 a fault occurs on L3 in..., then there are the following two possible situations for the misjudgment of the fault section: (1) The misjudged section is on the left side of the ranging node, such as Figure 5 L1 or L2 in...; (2) The misjudged section is on the right side of the ranging node, such as Figure 5 L4 or L5 in....

[0062] Therefore, to effectively verify the fault section location results, after the fault section location algorithm outputs the fault section, fault location is performed through the distance measurement nodes on both sides or one side of the fault section. For case (1), if the misjudged fault section is L1 or L2, then N1 and N3 are started for fault location. Since the actual fault point is located in L3, the fault power flows of N1 and N3 are both rightward. The distance measurement result of N1 exceeds the range of Type I line, while the distance measurement result of N3 is within the range of L3, indicating that the fault section judgment is incorrect and the actual fault section is L3. For case (2), if the misjudged fault section is L4 or L5, then N3 is started for fault location. Since the actual fault point is located in L3, the fault power flow of N3 is rightward, and its distance measurement result does not exceed the range of Type II line, but the distance measurement result does not match the range of the fault section location result, then it is determined that the fault section judgment is incorrect, and the fault section needs to be corrected according to the distance measurement result of N3 point.

[0063] (2) Two-way power flow system

[0064] As Figure 6 shown, this system is a two-way power flow system, and N1, N3, and N5 are distance measurement nodes. Assuming a fault occurs in L3, there are also the following two possible cases of misjudgment of the fault section: (1) The misjudged section is on the left side of the distance measurement node, such as Figure 6 L1 or L2 in Figure 6 ; (2) The misjudged section is on the right side of the distance measurement node, such as

[0065] L4 or L5 in

[0066] For case (1), if the misjudged fault section is L1 or L2, then N1 and N3 are started for fault location. Since the actual fault point is located in L3, the fault power flows of N1 and N3 are both rightward. The distance measurement result of N1 exceeds the range of Type I line, while the distance measurement result of N3 is within the range of L3, then it indicates that the fault section judgment is incorrect and the actual fault is located in L3. For case (2), if the misjudged fault section is L4 or L5, then N3 and N5 are started for fault location. Since the actual fault point is located in L3, the fault power flow of N3 is rightward and the fault power flow of N5 is leftward, then the distance measurement results of N3 and N5 do not exceed the range of Type II line, and the sum of the two results is approximately equal to the range of Type II line. After comprehensively considering the two results, it is determined that the fault point is located in L3 rather than L4 or L5, then it can be considered that the fault section judgment is incorrect and the actual fault is located in L3.

[0066] In addition, since fault location cannot guarantee zero error, in order to avoid the influence caused by ranging errors, the results of fault location should not be simply relied on. When the fault section location result of the distribution network is consistent with the verification result of fault location, since the possibility of misjudgment of both is relatively low, the fault location can be directly determined; however, when the fault section location result of the distribution network is inconsistent with the verification result of fault location, then the judgment sections of both and the intermediate sections need to be regarded as the objects to be investigated, and then the intersections between the sections should be focused on for investigation by means of line patrol.

[0067] Step 2: Based on the selected ranging nodes, obtain the three-phase fault signals of the sampling points and preprocess the three-phase fault signals.

[0068] Among them, the fault signals are obtained through a fault recorder. If the fault signals contain large noise, the fault signals also need to be denoised.

[0069] Step 3: Perform phase-mode transformation on the fault signals, obtain the mode components of the three-phase fault signals through Clarke transformation, and select appropriate mode components according to the fault type and selection rules.

[0070] In the actual ranging of the distribution network, due to the electromagnetic coupling between the three-phase lines, and when a fault occurs in the line, modal mixing will occur in each mode, and modal mixing will affect the extraction of the natural frequency. Therefore, it is necessary to decouple the sampling signals first. The phase-mode transformation adopted in this embodiment is Clarke transformation, and its phase-mode transformation form is:

[0071]

[0072] In the formula, I α , I β and I0 are the current α-mode component, β-mode component and 0-mode component respectively; I A , I B and I C are the phase current components of phase A, phase B and phase C respectively.

[0073] Among them, for different fault types, different moduli need to be selected, and the selection rules are shown in Table 1;

[0074] Table 1 Moduli corresponding to different fault types

[0075]

[0076] Step 4: Based on the selected mode components, use the IELM-VMD algorithm to perform modal decomposition on the signals. Take the VMD algorithm as the kernel, and alternately optimize the decomposition layer number K and the penalty factor α c through the double-layer structure of the IELM algorithm to obtain the IMF components.

[0077] Since the fault traveling wave is composed of the superposition of countless single signals with specific frequencies, there is a serious spectrum aliasing phenomenon. It is necessary to first decompose the original signal into several component signals by modal decomposition, and then perform subsequent operations on each component signal.

[0078] The VMD algorithm has excellent decomposition effects for non-linear and non-stationary signals, can effectively reduce phenomena such as endpoint effects, under-enveloping, over-enveloping, modal aliasing, and false components, and has extremely high decomposition accuracy for complex data. However, the VMD algorithm also has certain defects. It is sensitive to noise, and there is a modal aliasing phenomenon when the noise is severe. Moreover, the VMD algorithm requires manual setting of the decomposition layer number K and the penalty factor α c and other parameters, and their values extremely depend on the user's signal processing experience. And whether the values are appropriate directly affects the modal decomposition effect of the VMD algorithm. For example, the K value will affect the decomposition layer number of the VMD algorithm. If the K value is set too large, false IMF components will be generated. If it is set too small, the original signal will not be completely decomposed; and the penalty factor α c is related to the bandwidth of each IMF component. The larger α c , the smaller the bandwidth of the IMF component. The smaller α c , the larger the bandwidth. It can be seen that the VMD algorithm requires manual parameter setting, which not only makes the automation level of the algorithm too low, but also cannot guarantee the accuracy of modal decomposition, restricting its popularization and application in the field of distribution network fault location.

[0079] Aiming at the problem that the VMD algorithm overly depends on manual parameter setting, in this embodiment, the IELM-VMD algorithm is proposed based on the VMD algorithm. Its main idea is to use the envelope entropy of the IMF component as the evaluation function, and optimize and solve the parameters of the VMD through the IELM algorithm, so as to avoid manual setting of the parameters of the VMD algorithm, reduce the spectrum aliasing and over-decomposition phenomena caused by improper manual setting, and improve the automation level and accuracy of modal decomposition.

[0080] The process of constructing the IELM-VMD algorithm model includes:

[0081] (1) Construct a variational problem

[0082] After the original signal is decomposed by modal decomposition, several IMF components are obtained. Each IMF component should have a narrow bandwidth of the center frequency and make the estimated bandwidth of all IMFs have a minimum value.

[0083] And the estimation steps of the bandwidth of each IMF component are as follows:

[0084] 1) First, perform a Hilbert transform on the original signal X(t) to obtain each IMF component x k(t) analytical signal and single-sided spectrum;

[0085] 2) Then, shift the spectra of each IMF component to the corresponding fundamental frequency band, and estimate the center frequency e through hybrid estimation -jωkt ;

[0086] 3) Calculate the square of the 2-norm of the gradient of x k (t) to estimate the bandwidth of each IMF component.

[0087] If the original signal is decomposed into K IMF components, with the sum of the estimated bandwidths of each IMF component being minimized as the objective function and the superposition of each IMF component being equal to the original signal as the constraint, the following model can be established:

[0088]

[0089] In the formula, {x k (t)} is the set of K IMF components; {ω k} is the set of center frequencies of K IMF components; is the partial derivative, j is a complex number, t represents time, ω k represents the center frequency of the k-th IMF component, δ(t) is the Dirac generalized function; X(t) is the original signal.

[0090] (2) Solve the variational problem

[0091] To solve equation (2), it is necessary to introduce a penalty factor α c and the Lagrange operator λ, and transform the constrained optimization problem shown in equation (2) into an unconstrained optimization problem. The expression of the augmented Lagrangian function is:

[0092]

[0093] In the formula, are the IMF components at the (d + 1)-th iteration; are the center frequencies of the IMF components at the (d + 1)-th iteration; λ d+1 (t) is the Lagrange operator at the (d + 1)-th iteration. Solve the saddle point of equation (3) by the alternating direction method of multipliers, that is, the optimal solution of equation (3). Its basic idea is to alternately update λ from the d-th iteration to the (d + 1)-th iteration, and d+1 (t), while needs to be solved in the frequency domain to obtain each IMF component.

[0094] For iterative update, it is equivalent to the following formula:

[0095]

[0096] Assume and Solving Equation (4) in the frequency domain using Parseval's theorem, we can obtain:

[0097]

[0098] where represents the IMF component in the frequency domain, ω represents the frequency, refers to the signal in the frequency domain, represents the Lagrangian operator in the frequency domain, and d represents the number of iterations.

[0099] Replace ω in Equation (5) with ω - ω k , and the integral form in the non - negative frequency interval is:

[0100]

[0101] The solution obtained through quadratic optimization is:

[0102]

[0103] Since the central frequency ω k of x k (t) does not appear in the fidelity term of the reconstruction function and only exists in the bandwidth term. Therefore, for the iterative update, it can be equivalent to the following formula:

[0104]

[0105] Similarly, solving in the frequency domain, we have:

[0106]

[0107] For the update of λ d+1 , according to the dual ascent method of Equation (10), let t0 be the time step of dual ascent, we have:

[0108]

[0109] Set the convergence parameter ε, and stop the iterative update when the convergence criterion shown in Equation (11) is satisfied.

[0110]

[0111] After the iteration ends, perform the inverse Fourier transform on , and the real part of the obtained result is x k (t), while ω k is the frequency center of the mode function.

[0112] (3) Optimization of Decomposition Layers and Penalty Factor

[0113] The basis of the optimization problem is the evaluation function. Whether the evaluation function is reasonable determines whether the performance of the optimization algorithm is excellent. Based on the Shannon entropy, the concept of envelope entropy is proposed. Envelope entropy is a criterion that can evaluate the sparse characteristics of a signal. After Hilbert demodulation of each IMF component, the envelope signal b k can be obtained, and then through normalization processing by Equation (12), a probability distribution sequence

[0114]

[0115] The envelope entropy value B of this probability distribution sequence k reflects the uniformity of the signal distribution. When the regularity of the k-th IMF component is stronger, B k is smaller. The envelope entropy B k is calculated by the formula:

[0116]

[0117] In this implementation, the sum of the envelope entropies of each IMF component is used as the evaluation function, so the expression of the evaluation function is:

[0118]

[0119] Since the decomposition layer number K and the penalty factor α c have different characteristics, the value range of K is small and it is a discrete real number, while α c has a large value range and is a continuous real number. Therefore, it is more efficient to take an exhaustive approach for the value of K, while α c needs to be approximated and solved. The initial population of α c is:

[0120]

[0121] In the formula: Charge Q i is an individual of α c in the population; α c.ul , α c.ll are the upper and lower limit values of the value of α c respectively; Z is the population size. Equation (15) means to uniformly generate the initial population of the penalty factor α c within a reasonable range. The amount of charge q i carried by the charge Q i is calculated by the formula:

[0122]

[0123] In the formula: fV.i is the objective function value of the i-th charge in the population; f V.min is the minimum objective function value. The charge Q i The combined electric field force received in space is:

[0124]

[0125] In the formula: G is the optimal charge, and is f V.min The corresponding charge; the charge Q i The iteration formula of is:

[0126]

[0127] The convergence criterion is:

[0128]

[0129] In the formula: ε min is an arbitrarily small number.

[0130] Such as Figure 9 shown, the decomposition layer number K and the penalty factor α are alternately optimized through the double-layer structure of the IELM algorithm: c Perform alternating optimization:

[0131] In the first layer of the algorithm, the basic idea is to fix the penalty factor α c to the initial value, take the minimum value of the envelope entropy as the objective function, and solve the optimal decomposition layer number K through the IELM algorithm. The specific steps are as follows:

[0132] (1) Set the initial parameters. Fix the penalty factor α c to the initial value, set the initial value and the upper limit N of the decomposition layer number K V , and f V.min parameters such as the number of iterations D, the population size Z, etc.;

[0133] (2) Perform loop solution for . Solve all IMF components through equation (7) in a loop

[0134] (3) Perform loop solution for . Solve the central frequencies of all IMF components through equation (9) in a loop

[0135] (4) Solve for λ d+1 . Solve for λ according to equation (10) d+1 , d = d + 1;

[0136] (5) Determine whether the end condition is satisfied. When the formula (11) is satisfied, the iteration ends, and at this time, K IMF components can be obtained; otherwise, execute step (2).

[0137] (6) Calculate the envelope entropy and the objective function value. Calculate the envelope entropy B of each obtained IMF component through formulas (12) and (13) respectively k , and calculate the evaluation function value f according to formula (14) V . When f V < f V.min , let f V.min = f V , K best = K;

[0138] (7) Find the optimal K value. Let K = K + 1. When K ≤ N VMD , let k = 1, d = 1 and execute step (2). When K > N V , K = K best , and perform step (8);

[0139] In the second layer of the algorithm, the basic idea is to regard the optimal decomposition layer number K as a fixed value, take the minimum value of the envelope entropy as the objective function, and solve the optimal penalty factor α through the IELM algorithm c , which specifically includes the following steps:

[0140] (8) Generate the initial population of the penalty factor α c . Generate the initial population of α c according to formula (15), and initialize each parameter;

[0141] (9) Obtain the IMF components. Substitute the K value and α c.i into steps (2) to (5) to obtain each IMF component;

[0142] (10) Calculate the envelope entropy and the objective function value. Calculate the envelope entropy B of each obtained IMF component through formulas (12) and (13) respectively k , and calculate the objective function value f corresponding to each charge in the population according to formula (14) V.i , find the minimum value f V.min , and regard the corresponding charge as the optimal charge G, and record the corresponding set of IMF components {x k (t)} best ;

[0143] (11) Calculate the charge amount of each element in the population. Calculate the charge amount q of each charge in the population according to formula (16) i ;

[0144] (12) Calculate the forces on each element in the population. Calculate the combined force F on each charge in the population according to Equation (17). i ;

[0145] (13) Iterate the charges. Iterate the charges in the population according to Equation (18);

[0146] (14) Determine whether the convergence condition is satisfied. Continue to execute steps (9) - (10) on the iterated population to obtain f V.min , and check whether the convergence condition shown in Equation (19) is satisfied. If the convergence condition is not satisfied, let d = d + 1 and continue to execute step (11); if the convergence condition is satisfied, output {x k (t)} best .

[0147] Obtain the optimal parameters required for modal decomposition through an optimization algorithm, with the expectation that the modal decomposition achieves the best effect.

[0148] Step 5: Based on the obtained IMF components, extract the natural frequencies of the IMF components.

[0149] It is difficult to directly obtain the natural frequencies of the IMF components in the actual distribution network. It is necessary to indirectly obtain the natural frequencies of the traveling waves through other means. If you want to obtain the natural frequencies of the signal, you need to convert the signal from the time domain to the frequency domain. Commonly used time-domain to frequency-domain conversion methods include fast Fourier transform, wavelet transform, multiple signal classification, etc.

[0150] Regarding the specific principles of the fast Fourier transform and wavelet transform, the specific content has been publicly disclosed in the prior art, and will not be elaborated in this embodiment.

[0151] In this embodiment, the multiple signal classification method is selected to extract the natural frequencies of the IMF components, which specifically includes:

[0152] Multiple Signal Classification (MUSIC) is a classic spatial direction estimation method. This method first performs eigenvalue decomposition on the covariance matrix of the sampled signal to establish the signal subspace and the noise-containing subspace, thereby establishing a spatial spectrum function using the orthogonality of the two subspaces, and finally obtaining the distance and azimuth information of the target by locating the spectral peak.

[0153] Although the natural frequency of the fault traveling wave is a statistically time-varying signal, the frequency value does not change with time. Only the amplitude of the frequency decays exponentially with time, while the system harmonics and noise hardly decay with time. Based on this characteristic, the natural frequency of the traveling wave can be distinguished from other interference signals. Let f kdenotes the natural frequency of the sampling signal. Therefore, the sampling signal is composed of the superposition of the natural frequency and white noise. The model of the fault traveling wave is as follows:

[0154]

[0155] where: a k and f k are respectively the amplitude and frequency of the k-th complex sinusoidal signal; v(p) is white noise; p is the number of sampling points. Rewrite Equation (20) in matrix form:

[0156]

[0157] where: M is the length of the time window vector. The autocorrelation function of the fault traveling wave X(p) is:

[0158]

[0159] where: E is the identity matrix. Decompose Equation (22) in terms of eigenvalues in the form of:

[0160]

[0161] where: sort λ m such that λ1 ≥ λ2 ≥ … ≥ λ m ; q m is its corresponding eigenvector. Obtain the eigenvectors of the white noise and the sampling signal. Due to the orthogonality between the noise subspace and the signal subspace, the spatial spectrum obtained by the MUSIC method is:

[0162]

[0163] where: the polynomial has M - 1 roots, among which there are P complex exponentials corresponding to the signals, which are the same for all M - P noise eigenvectors, and the rest are different. To reduce the peak value of the noise, the average value of the spectra of all M - P noise eigenvectors can be obtained through Equation (25).

[0164]

[0165] The P-th peak value of the power spectrum of Equation (25) corresponds to the frequency components in the signal, that is, the power spectrum of MUSIC.

[0166] The advantages of the above scheme are that the MUSIC algorithm has the advantages of high precision, good flexibility, wide frequency domain, etc., and can also identify complex exponential signals with close frequencies well. Compared with the fast Fourier transform and wavelet transform, it has better accuracy and noise immunity in extracting the natural frequency of the fault traveling wave, and has high precision even for short signal lengths and fast attenuation speed of fault transient traveling waves.

[0167] Step 6: Calculate the fault distance based on the relationship between the natural frequency of the IMF component and the fault distance.

[0168] In the ideal open - circuit or short - circuit state, the relationship between the natural frequency of the fault transient traveling wave, the fault distance, and the wave velocity is:

[0169]

[0170] Where: f is the natural frequency; x is the fault distance; c is the wave velocity. However, in an actual distribution network, the ideal open - circuit or short - circuit state does not exist. The natural frequency of the fault traveling wave is related not only to the fault distance but also to the reflection characteristics of the line boundary. In this embodiment, taking the Figure 7 shown line fault model as an example, the relationship between relevant parameters is explained.

[0171] If Figure 7 a fault occurs at point A in Figure 7 , to obtain the reflection characteristics at the line boundary (i.e., between bus 1 and bus 2), the line fault model in Figure 8 is equivalent to the two - port network shown in

[0172] The reflection characteristics of its boundary are described by the characteristic impedance and the controlled source, and the input - state - output model is used to describe the traveling - wave ranging problem. Figure 8 It can be known from

[0173]

[0174] Where: U1(t) and U2(t) are the terminal voltages at both ends of the line respectively; G1(t) and G2(t) are the voltage sources of the systems on both sides of the line respectively; Z1 and Z2 are the equivalent impedances of the systems on both sides of the line respectively; Z C is the characteristic impedance of the line; W1(t) and W2(t) are the equivalent controlled sources (representing state variables) respectively. Since frequency affects the calculation of impedance, when the equivalent impedance or characteristic impedance is not a pure resistor, it is more convenient to analyze Equation (27) through Laplace transform. After Laplace transform, Equation (27) becomes:

[0175]

[0176] Where: P(s) is the Laplace - domain operator, representing the line time delay, with P(s)=e -sT ; Γ1 and Γ2 are the reflection coefficients at the line boundary in the Laplace domain, and their values are given by Equation (29).

[0177]

[0178] So far, the expression derivation of the state variables is completed, and theoretically the solutions of the terminal current and voltage at any time can be obtained. For example, if the sampling point is set at bus 1, after a fault occurs, the voltage signal collected at the sampling point is defined by Equation (28), and its spectrum consists of the spectra of G1(s) and W1(s). The frequency of G1(s) is the power frequency under ideal conditions, while the spectrum of W1(s) is the natural frequency of the fault traveling wave, and its spectrum is the root of Equation (30).

[0179] 1 - Γ1(s)Γ2(s)P 2 (s) = 0 (30)

[0180] Assume that the reflection coefficients Γ1 and Γ2 of bus 1 and bus 2 are both real numbers, and there is a characteristic equation:

[0181] 1 - Γ1Γ2e -2fT = 0 (31)

[0182] Where: T is the time for the traveling wave to move from the fault point to the sampling point. Through e jθ = sinθ + cosθ, Equation (31) is transformed into an exponential form:

[0183] e 2fT = Γ1Γ2e j2kπ (32)

[0184] Where: k = 0, ±1, ±2, ……. Further solving Equation (32), we get:

[0185]

[0186] Where: f k is the frequency of each component in the natural frequency. However, in the actual distribution network, its boundary conditions are relatively complex and may contain components related to ω such as inductance and capacitance. Therefore, it is difficult to obtain an analytical solution in the form of Equation (33). But if the main components of the natural frequency can be obtained in other ways, the fault distance x f can be obtained by analyzing the characteristic Equation (30). Let the Laplace transform coefficient s = σ + jω, and substitute ω k = 2πf k , T = x f / c k , into Equation (30). If it is assumed that both ends of the line are in short circuit, open circuit or pure resistance state, that is, Γ1 and Γ2 are both real numbers, we have:

[0187]

[0188] Where: c k is the wave velocity corresponding to f k . When an inductor or capacitor is connected to one end of the line, the reflection coefficient at that end becomes complex, and its expression is:

[0189]

[0190] Where: θ1 and θ2 are the reflection angles of the reflection coefficients at both ends, satisfying -π / 2 < θ1 + θ2 < π / 2. From the equality of the imaginary parts of Equation (35), we can obtain:

[0191]

[0192] Equations (34) and (36) are the calculation formulas for the fault distance when the reflection coefficient is real and complex, respectively. When f k is the main component in the spectrum, k takes 0; when f k is the second-highest component in the spectrum, k takes 1; and so on for the rest of the cases.

[0193] Among them, calculating parameters such as line parameters, wave velocity, reflection angle of the sampling point, and reflection angle of the fault point based on the natural frequency of the IMF component includes:

[0194] (1) Calculation of line parameters

[0195] The calculation formula for the characteristic impedance matrix is:

[0196]

[0197] Where: Z s and Y s are the modal impedance matrix and modal admittance matrix at the main frequency of the sampling signal, respectively.

[0198] (2) The expression for the velocity of the traveling wave is:

[0199]

[0200] The modulus wave velocity matrix c m includes the wave velocities of each modal component, and the modulus can be selected according to Table 1 for different fault types.

[0201] (3) The reflection coefficients Γ1 and Γ of the signal sampling point and the fault point f are respectively:

[0202]

[0203] Where: Z1 is the system equivalent impedance matrix; T is the Clarke transformation matrix. When calculating the reflection coefficient Γ of the fault point f , the influence of different fault types on Γ f also needs to be considered. Assuming that phase A is the special phase and the transition resistance is R g , there are the following four cases.

[0204] For single-phase grounding:

[0205]

[0206] Where: Z mα and Z m0 are the characteristic impedances under the 0-mode and α-mode of the line respectively.

[0207] During a two-phase short circuit:

[0208]

[0209] Where: Z mβ is the characteristic impedance under the β-mode of the line.

[0210] During a two-phase grounding short circuit:

[0211]

[0212] During a three-phase short circuit:

[0213]

[0214] Calculate parameters such as line parameters, wave velocity, reflection angle of the sampling point, and reflection angle of the fault point, and finally calculate the fault distance through Equation (34) or Equation (36).

[0215] Case study:

[0216] To verify the effectiveness of the proposed distribution network fault location method based on the IELM-VMD algorithm in this invention, this invention takes F4-1 of the IEEE 33-node distribution network as an example, compares and tests the proposed IELM-VMD algorithm in this invention with the current mainstream VMD and EMD algorithms, and demonstrates and analyzes the modal decomposition effect and ranging accuracy of the proposed algorithm in this invention; takes F4-4 of the IEEE 33-node distribution network as an example, and demonstrates and analyzes the verification mechanism of the proposed algorithm in this invention.

[0217] To analyze the fault traveling wave more accurately, this invention changes the parameters of the transmission line in the example from lumped parameters to distributed parameters, and sets two types of transmission lines, namely 300 and 301, based on the provided standard parameters of overhead lines and cables. This invention builds an example model and tests the proposed algorithm.

[0218] Analysis of modal decomposition effect:

[0219] Taking F4-1 as an example, the specific parameter settings are as follows: the total simulation duration is 0.3 s, and a three-phase metallic short circuit occurs on L 10 at 0.02 s, the fault point is located 4 km to the right of node N 10 , and the transition resistance is 0. Since the fault section location result is L 10, according to the verification mechanism, the ranging node is selected as N 10 , and the sampling frequency is set to 1 MHz.

[0220] First, the β-mode component of the fault current signal is decomposed by the EMD algorithm. Since the EMD algorithm does not require manual parameter setting, the original signal can be directly decomposed by EMD. Arrange the decomposed IMF components in descending order of frequency, and the decomposition effect is as Figure 10 shown.

[0221] By Figure 10 it can be seen that the characteristic information contained in the IMF1 and IMF2 components is relatively rich, while the characteristic information contained in the remaining IMF components is less. This is because the characteristic information after being processed by the EMD algorithm will be mainly concentrated in the high-frequency IMF components, and this characteristic will lead to the easy occurrence of mode mixing phenomenon in the decomposed high-frequency IMF components.

[0222] Secondly, the β-mode component of the fault current signal is decomposed by the VMD algorithm. According to experience, the decomposition layer number K is set to 10, and the penalty factor α c is 1500. Arrange the IMF components in descending order of frequency, and the decomposition effect is as Figure 11 shown. Since the line is short, the natural frequency of the traveling wave is relatively high. For easy observation, the high-frequency IMF components of the present invention are enlarged and displayed.

[0223] By Figure 11 it can be seen that the VMD algorithm decomposes the original signal into 10 IMF components, and the frequency decreases in turn. Among them, the frequency of IMF10 is the lowest and is the same as the fundamental frequency of the original signal, while the frequency of IMF1 is the highest. According to the rule, IMF2 should have the frequency second only to IMF1, but by Figure 11 it can be seen that IMF2 does not have a frequency performance, indicating that over-decomposition occurs in this decomposition, which means that the values of the decomposition layer number K and the penalty factor α c in this decomposition are unreasonable.

[0224] Finally, the β-mode component of the fault current signal is decomposed by the IELM-VMD algorithm, and the decomposition effect is as Figure 12 shown. The optimal decomposition layer number K obtained is 8, and the optimal penalty factor α c is 1281.5.

[0225] Still, the high-frequency IMF components are enlarged and displayed. By Figure 12It can be seen that there is no over-decomposition phenomenon in each IMF component, and there is almost no mode mixing. The decomposition effect is better than that of the VMD algorithm that simply relies on empirical parameter settings. This indicates that the proposed IELM-VMD algorithm in this paper can not only save the steps of manual parameter setting, improve the automation level of the algorithm, reduce the usage threshold of the algorithm, but also reduce the over-decomposition and mode mixing phenomena caused by improper manual parameter setting, and improve the mode decomposition accuracy.

[0226] Range measurement accuracy verification:

[0227] This invention still takes F4-1 as an example, and also takes the VMD and EMD algorithms as comparison algorithms to verify the range measurement accuracy of the proposed distribution network fault location method based on IELM-VMD.

[0228] First, the detailed calculation process of the proposed algorithm in this invention is demonstrated.

[0229] This invention selects the MUSIC algorithm to extract the natural frequencies of each IMF component. Since the main frequency component of IMF8 is the fundamental frequency, this paper only extracts the natural frequencies of IMF1 to IMF7 and screens out the frequency with the largest energy among them. The extraction results are as Figure 13 shown.

[0230] From Figure 13 it can be known that the main frequency f1 of the β-mode signal of the fault current is 35762 Hz. Since the line L 10 is a 300-type line, the parameters per unit length of the 300-type line are: R x = 0.17 Ω / km, L x = 1.21×10 -3 H / km, C x = 0.0097×10 -6 F / km. Substitute the line parameters and the main frequency f1 into Equation (37) to obtain the characteristic impedance matrix Z m as:

[0231]

[0232] Since this line is a lossy line, the reflection coefficients Γ1 and Γ f at the sampling point and the fault point are both complex matrices. Among them, the sampling point reflection coefficient Γ1 is calculated by Equation (39), and its calculation result is:

[0233]

[0234] Since the fault type is three-phase short circuit, the reflection coefficient Γ f at the fault point is calculated by Equation (43), and its calculation result is:

[0235]

[0236] Select the β-mode components of Γ1 and Γ f to obtain the reflection angles θ1 and θ between the sampling point and the fault point f both being π. Then substitute the main frequency f1 into Equation (38) to obtain the wave velocity corresponding to the main frequency f1 as:

[0237]

[0238] Select its β-mode component as the wave velocity for calculation. Therefore, the wave velocity corresponding to the main frequency f1 is c1 = 2.9189×10 5 km / s. Finally, substitute the parameters obtained above into Equation (36) to obtain the fault distance x f as:

[0239]

[0240] It shows that the fault point is located 4.081 km to the right of N 30 with a difference of only 0.081 km from the set parameter of the fault distance.

[0241] Change the transition resistance and fault type, and compare and test the algorithm proposed in this invention with the VMD algorithm. The parameters of the VMD algorithm are uniformly set as K = 10, α c = 1500, and the ranging node is uniformly N 10 . The comparison results are shown in Table 2.

[0242] Table 2 Fault ranging simulation results based on example F4-1

[0243]

[0244]

[0245] Table 2 (continued)

[0246]

[0247] As can be seen from Table 2, the change in the fault type has little effect on the ranging accuracy of the three algorithms, while the transition resistance has a greater impact on the ranging accuracy. This is because the transition resistance affects the reflection coefficient at the fault point, resulting in a decrease in the ranging accuracy. For the EMD and VMD algorithms, their ranging accuracies under various transition resistances are similar, but the overall ranging accuracy of the VMD algorithm is slightly better than that of the EMD algorithm, indicating that the decomposition effect of the VMD algorithm on the original signal is better than that of the adaptive algorithm without manual parameter setting. For the proposed IELM-VMD algorithm of the present invention, when the fault is a metallic short circuit, the proposed IELM-VMD algorithm of the present invention has an error of approximately 1.5% - 2% under various fault types, and the ranging accuracy is improved by approximately 2.2 - 2.8 times compared with the VMD algorithm; when the transition resistance is 10Ω, the proposed IELM-VMD algorithm of the present invention has an error of approximately 2.5% - 3.2% under various fault types, and the ranging accuracy is improved by approximately 2 - 2.3 times compared with the VMD algorithm; when the transition resistance is 100Ω, the proposed IELM-VMD algorithm of the present invention has an error of approximately 5% - 6% under various fault types, and the ranging accuracy is improved by approximately 1.3 - 1.7 times compared with the VMD algorithm.

[0248] Thus, it can be seen that the proposed algorithm of the present invention has good ranging accuracy under various transition impedances and fault types, and compared with the VMD algorithm, the proposed algorithm of the present invention can effectively improve the ranging accuracy under various transition impedances, and the smaller the transition impedance, the more obvious the improvement effect.

[0249] Verification of the effectiveness of the verification mechanism:

[0250] Based on the fault section location result, taking F4-4 as an example, the verification mechanism of the proposed distribution network fault ranging method based on IELM-VMD of the present invention is verified. The enlarged topology diagram of the distribution network of example F4-4 is as Figure 14 shown.

[0251] In example F4-4, the fault occurs in section L 31 , and the information uploaded by the FTU of node N 31 is distorted from 1 to 0. The fault section location result of the distribution network is section L 30 . According to the verification mechanism, the ranging node should be selected as node N 30 . From the above conclusions, it can be seen that the fault type has little effect on the ranging accuracy of the fault ranging. Therefore, the influence of the fault type on the verification mechanism is ignored in the test, and the fault type is uniformly set to a single-phase grounding fault. The verification mechanism of the proposed algorithm of the present invention is simulated and tested through different fault distances and transition resistances. The test results are shown in Table 3.

[0252] As can be seen from Table 3, when the fault point location is far from the end boundary of L 31 , the sections to be investigated are L 30 and L31 , according to the rule for selecting the section to be investigated in Section S5, the verification result of the fault location is section L 31 , indicating that the fault location effectively verifies the location result of the fault section. However, when the fault point is close to the end boundary of L 31 , the sections to be investigated obtained are L 30 , L 31 and L 32 , indicating that the verification result of the fault location is L 32 . This is because there is a positive error in the algorithm proposed in the present invention, and the distance obtained by the fault location crosses the node N 31 and falls within the section L 32 . It shows that there is a verification dead zone in the fault section verification mechanism proposed in the present invention. The ranges of the verification dead zones under the three types of transition impedances are 1.2%, 2% and 4.5% respectively, all within the acceptable range. The junction between the sections to be investigated can be key investigated by means of line patrol.

[0253] Table 3 Simulation test results of the fault section verification mechanism for the distribution network

[0254]

[0255]

[0256] Embodiment 2

[0257] This embodiment provides a distribution network fault location system based on the IELM-VMD algorithm, including:

[0258] A ranging node selection module, configured to determine ranging nodes based on the fault section location result and the verification mechanism;

[0259] A fault signal acquisition module, configured to acquire three-phase fault signals of sampling points based on the selected ranging nodes;

[0260] A mode component acquisition module, configured to perform phase-mode transformation on the three-phase fault signals, obtain the mode components of the three-phase fault signals through Clarke transformation, and select mode components according to the fault type and selection rules;

[0261] A modal decomposition module, configured to perform modal decomposition on the signal based on the selected mode components by using the IELM-VMD algorithm, use the VMD algorithm as the kernel, use the envelope entropy of the IMF components as the evaluation function, and optimize and solve the decomposition layer number and penalty factor of the VMD through the double-layer structure of the IELM algorithm to obtain IMF components;

[0262] A fault location module, configured to extract the natural frequencies of the IMF components based on the IMF components, and calculate the fault distance based on the relationship between the natural frequencies of the IMF components and the fault distance.

[0263] Embodiment III

[0264] This embodiment provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the steps in the distribution network fault location method based on the IELM-VMD algorithm as described in Embodiment I above.

[0265] Embodiment IV

[0266] This embodiment provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the steps in the distribution network fault location method based on the IELM-VMD algorithm as described in Embodiment I above.

[0267] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention may have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A fault location method for distribution networks based on the IELM-VMD algorithm, characterized in that Including: Determine the ranging nodes based on the fault section location result; According to the fault section location result of the distribution network, take the connection nodes between lines with different parameters and the nodes near the power source side as the ranging nodes; Based on the selected ranging nodes, obtain the three-phase fault signals of the sampling points; Perform phase-mode transformation on the three-phase fault signals, obtain the modal components of the three-phase fault signals through Clarke transformation, and select the modal components according to the fault type and selection rules; Based on the selected modal components, use the IELM-VMD algorithm to perform modal decomposition on the signals. Take the VMD algorithm as the kernel, use the envelope entropy of the IMF components as the evaluation function, and optimize and solve the decomposition layer number and penalty factor of VMD through the double-layer structure of the IELM algorithm to obtain the IMF components; Based on the IMF components, extract the inherent frequencies of the IMF components, and calculate the fault distance based on the relationship between the inherent frequencies of the IMF components and the fault distance; The construction process of using the envelope entropy of the IMF components as the evaluation function includes: taking the minimum sum of the estimated bandwidths of each IMF component as the objective function, and taking the superposition of each IMF component equal to the original signal as the constraint to establish a constrained optimization problem; Introduce the penalty factor and the Lagrange operator to transform the constrained optimization problem into an unconstrained optimization problem; solve the unconstrained optimization problem through the alternating direction multiplier method, and use the idea of alternating update to obtain each IMF component; after demodulating each IMF component, obtain the envelope signal, normalize the envelope signal to obtain the probability distribution sequence, obtain the envelope entropy based on the probability distribution sequence, and take the sum of the envelope entropies of each IMF component as the evaluation function; The decomposition level of the double-layer structure by the IELM algorithm K and the penalty factor are optimized and solved. The specific steps are as follows: In the first layer of the IELM algorithm, the penalty factor is fixed to the initial value, and the minimum envelope entropy is taken as the objective function, and the optimal decomposition level is solved by the IELM algorithm K； In the second layer of the IELM algorithm, the optimal decomposition level K is regarded as a fixed value, and the minimum envelope entropy is taken as the objective function, and the optimal penalty factor is solved by the IELM algorithm .

2. The method for fault location in a distribution network based on the IELM-VMD algorithm according to claim 1, characterized in that, If the original signal is decomposed into K IMF components , taking the minimum sum of the estimated bandwidths of each IMF component as the objective function and the superposition of each IMF component being equal to the original signal as the constraint, the following model can be established: wherein, is the set of K IMF components; is the set of the central frequencies of the K IMF components; is the partial derivative, j is a complex number, t represents time, represents the central frequency of the k-th IMF component, is the Dirac generalized function, is the original signal.

3. The fault location method for a distribution network based on the IELM-VMD algorithm according to claim 1, characterized in that The expression of the unconstrained optimization problem is: In the formula, is K a set of IMF components, is K a set of central frequencies of the IMF components, is the Lagrange operator, is the penalty factor, is the Dirac generalized function; is the original signal.

4. The fault location method for a distribution network based on the IELM-VMD algorithm according to claim 1, characterized in that, Based on the IMF components, the method for extracting the inherent frequencies of the IMF components uses the multiple signal classification estimation method.

5. A distribution network fault location system based on the IELM-VMD algorithm, characterized in that, Including: A ranging node selection module, used to determine the ranging nodes based on the fault section location result and the verification mechanism; According to the fault section location result of the distribution network, take the connection nodes between lines with different parameters and the nodes near the power source side as the ranging nodes; A fault signal acquisition module, used to obtain the three-phase fault signals of the sampling points based on the selected ranging nodes; A modal component acquisition module, used to perform phase-mode transformation on the three-phase fault signals, obtain the modal components of the three-phase fault signals through Clarke transformation, and select the modal components according to the fault type and selection rules; A modal decomposition module, used to perform modal decomposition on the signals based on the selected modal components using the IELM-VMD algorithm. Take the VMD algorithm as the kernel, use the envelope entropy of the IMF components as the evaluation function, and optimize and solve the decomposition layer number and penalty factor of VMD through the double-layer structure of the IELM algorithm to obtain the IMF components; A fault ranging module, used to extract the inherent frequencies of the IMF components based on the IMF components, and calculate the fault distance based on the relationship between the inherent frequencies of the IMF components and the fault distance; The construction process of using the envelope entropy of the IMF components as the evaluation function includes: taking the minimum sum of the estimated bandwidths of each IMF component as the objective function, and taking the superposition of each IMF component equal to the original signal as the constraint to establish a constrained optimization problem; Introduce the penalty factor and the Lagrange operator to transform the constrained optimization problem into an unconstrained optimization problem; solve the unconstrained optimization problem by the alternating direction method of multipliers, and obtain each IMF component by using the idea of alternating update; after demodulating each IMF component, obtain the envelope signal, normalize the envelope signal to obtain the probability distribution sequence, obtain the envelope entropy based on the probability distribution sequence, and take the sum of the envelope entropies of each IMF component as the evaluation function. The decomposition level by the double-layer structure of the IELM algorithm K and the penalty factor are optimized and solved. The specific steps are as follows: In the first layer of the IELM algorithm, the penalty factor is fixed as the initial value, and the minimum envelope entropy is taken as the objective function, and the optimal decomposition level is solved by the IELM algorithm K； In the second layer of the IELM algorithm, the optimal decomposition level K is regarded as a fixed value, and the minimum envelope entropy is taken as the objective function, and the optimal penalty factor is solved by the IELM algorithm .

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the distribution network fault location method based on the IELM-VMD algorithm described in any one of claims 1-4.

7. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the distribution network fault location method based on the IELM-VMD algorithm described in any one of claims 1-4.

Images