Evolution traceability analysis method and system for power distribution network coefficient complex fault

Abstract

The invention discloses a power distribution network coefficient complex fault evolution traceability analysis method and system, and the method comprises the steps: obtaining a first signal of a target power distribution line, and carrying out the first preprocessing of the first signal, and obtaining a second signal; establishing a first characteristic matrix, and obtaining a third signal according to the second signal; and inputting the third signal into a pre-trained first judgment network, obtaining the fault category of the fault point, and performing fault position evolution traceability. Through time reversal operation, the focusing performance of a power grid fault signal at a fault point is remarkably enhanced, in combination with BLT hypermatrix modeling, amplitude attenuation, phase change and energy reflection characteristics of signal propagation in a complex power grid are accurately described, and through deep feature learning of a Siamese network, in combination with energy focusing verification and least square optimization, the power grid fault signal is obtained. And the fault classification and position precision in a complex fault scene is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power system fault diagnosis, and in particular to an evolution tracing and source analysis method and system for complex faults in a distribution network coefficient. Background Art

[0002] With the rapid development of modern power systems, the scale and complexity of distribution networks are constantly increasing. The occurrence of various faults has put forward higher requirements for the safety and reliability of power grids. During the operation of distribution networks, fault events often occur due to various factors such as line aging, external environmental influences, and equipment defects. Especially in distribution networks with complex topologies, faults are often diverse and complex, such as multi-point short circuits, grounding faults, and the superposition of multiple faults. Traditional fault detection and location methods, such as fault detection based on the action logic of protection devices, have good effects in simple fault scenarios, but often have certain limitations in the diagnosis and location of complex faults. In recent years, time domain reflectometry (TDR), frequency domain reflectometry (FDR), and fault detection methods based on artificial intelligence have gradually been applied to fault location in distribution networks. However, these methods still face significant technical challenges in processing multi-dimensional signals, dynamically changing fault conditions, and precise positioning in complex environments.

[0003] The existing related technologies mainly face the following shortcomings: traditional methods based on protection logic and simple signal analysis lack comprehensive modeling capabilities for complex multi-point faults and dynamic fault evolution paths, resulting in insufficient positioning accuracy, especially when multiple fault points affect each other, which is prone to misjudgment. Although time domain reflection and frequency domain reflection technologies can analyze fault signals, they do not handle multi-path reflection, superposition and attenuation of signal propagation in complex power grids well, resulting in difficulty in signal feature extraction. Although artificial intelligence-based analysis methods have strong capabilities in pattern recognition, they rely on large-scale training samples and have weak ability to distinguish complex fault types under conditions of few samples, especially in terms of positioning accuracy and fault classification effects, which are difficult to meet actual needs. Current technologies still have a lot of room for improvement in high-precision fault location, fault category discrimination and multi-source signal processing in complex power grid environments. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the above existing problems, the present invention is proposed.

[0006] Therefore, the present invention provides a method and system for tracing the evolution of complex faults in distribution network coefficients, which can solve the problems mentioned in the background technology.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, the present invention provides a method for tracing the evolution of complex faults in a distribution network, comprising:

[0009] Acquire a first signal of a target power distribution line, and perform a first preprocessing on the first signal to obtain a second signal;

[0010] Establishing a first characteristic matrix, and obtaining a third signal according to the second signal;

[0011] The third signal is input into the pre-trained first judgment network to obtain the fault category of the fault point and to trace the evolution of the fault location.

[0012] As a preferred solution of the evolution tracing and analysis method of complex faults of distribution network coefficients according to the present invention, wherein: the establishment of the first characteristic matrix and the acquisition of the third signal according to the second signal include:

[0013] The first characteristic matrix is ​​an arbitrary matrix simulating the propagation characteristics of the voltage and current at the fault point;

[0014] The second signal is a signal obtained by performing a first preprocessing on the first signal collected at a plurality of collection points.

[0015] As a preferred solution of the method for tracing the evolution of complex faults in distribution network coefficients described in the present invention, the step of inputting the third signal into a pre-trained first judgment network, obtaining the fault category of the fault point, and tracing the evolution of the fault location includes:

[0016] The input of the first judgment network is the third signal, and the output is the probability distribution of the fault category of the fault point or any network that can directly or indirectly solve the relevant parameters of the probability distribution of the fault category;

[0017] The performing fault location evolution tracing includes performing fault location evolution tracing by using a preset first evolution tracing strategy.

[0018] As a preferred solution of the evolutionary source tracing analysis method for complex faults of distribution network coefficients described in the present invention, the first evolutionary source tracing strategy includes:

[0019] After determining the fault type of the fault point, simulating the current values ​​of different fault points based on the first characteristic matrix, and establishing a first fault judgment set according to the current values;

[0020] The first fault judgment set is a data set sorted according to the magnitude of current energy;

[0021] A first frequency domain energy threshold is preset, and the first fault judgment set is judged one by one in a fixed order based on the first frequency domain energy threshold.

[0022] As a preferred solution of the evolutionary source tracing analysis method for complex faults of distribution network coefficients described in the present invention, the first evolutionary source tracing strategy also includes:

[0023] If the frequency domain energy of the fault point in the first fault judgment set exceeds the set energy threshold, it is recorded as a reasonable fault point;

[0024] Otherwise, the first fault judgment set is judged one by one in a fixed order.

[0025] As a preferred solution of the evolutionary source tracing analysis method for complex faults of distribution network coefficients described in the present invention, the first evolutionary source tracing strategy also includes:

[0026] According to the distance between the reasonable fault point and the actually observed fault point and the signal propagation speed, the time deviation is converted into a distance deviation by using the difference between the theoretical propagation time of the signal and the actually observed time;

[0027] The coordinates of the reasonable fault point are adjusted according to the distance deviation, and the optimization target is constructed using the least squares method. The sum of the squares of the deviations between the reasonable fault point response of the corrected point and the actual observed response is minimized. The final fault point position is obtained through iterative optimization using the gradient descent method.

[0028] As a preferred solution of the evolution tracing analysis method of complex faults of distribution network coefficients described in the present invention, the fault location evolution tracing includes: updating the first characteristic matrix by taking the final fault point location as the reasonable fault point location.

[0029] In a second aspect, the present invention provides an evolution tracing and analysis system for complex faults in distribution network coefficients, comprising:

[0030] A signal acquisition module, used for acquiring a first signal of a target power distribution line, and performing a first preprocessing on the first signal to obtain a second signal;

[0031] A matrix establishment module, used to establish a first characteristic matrix and obtain a third signal according to the second signal;

[0032] The identification module is used to input the third signal into the pre-trained first judgment network, obtain the fault category of the fault point, and trace the evolution of the fault location.

[0033] In a third aspect, the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned method when executing the computer program.

[0034] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the method described above when executed by a processor.

[0035] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention proposes an evolution tracing analysis method and system for complex faults in distribution network coefficients, obtains a first signal of the target distribution line, and performs a first preprocessing on the first signal to obtain a second signal; establishes a first characteristic matrix, and obtains a third signal according to the second signal; inputs the third signal into the pre-trained first judgment network to obtain the fault category of the fault point, and performs evolution tracing of the fault location. Through the time reversal operation, the focusing of the power grid fault signal at the fault point is significantly enhanced. Combined with the BLT super-matrix modeling, the amplitude attenuation, phase change and energy reflection characteristics of signal propagation in complex power grids are accurately described. Through the deep feature learning of the Siamese network, combined with energy focusing verification and least squares optimization, the fault classification and location accuracy in complex fault scenarios are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0037] Figure 1 A method flow chart of an evolution tracing and source analysis method and system for complex faults in distribution network coefficients provided by an embodiment of the present invention;

[0038] Figure 2 A schematic diagram of a flow chart of a signal verification of a method for tracing the source of a complex fault in a distribution network coefficient and a system provided by an embodiment of the present invention;

[0039] Figure 3 An internal structure diagram of a computer device of a method and system for tracing the evolution of complex faults in a distribution network coefficient provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0041] Example 1

[0042] Reference Figure 1-Figure 3 , which is the first embodiment of the present invention, and provides a method and system for tracing the evolution of complex faults in distribution network coefficients, including:

[0043] There are some shortcomings in the existing related technologies. For example, traditional fault detection methods often rely on fixed protection logic, which can easily lead to misjudgment and inaccurate positioning when facing complex faults. In addition, although time domain reflection and frequency domain reflection technologies have certain advantages in signal analysis, they still face challenges when dealing with multipath reflection, signal superposition and attenuation in complex power grids. Although the analysis method based on artificial intelligence performs well in pattern recognition, it has weak ability to distinguish complex fault types under the condition of few samples, especially in terms of positioning accuracy and fault classification effect, it is difficult to meet actual needs.

[0044] The present application provides a method that can effectively solve the above-mentioned problems. Next, a method for tracing the evolution of complex faults in the distribution network coefficient will be described in detail in combination with multiple embodiments.

[0045] Figure 1 A method flow chart of an evolution tracing analysis method and system for complex faults of distribution network coefficients is shown, including:

[0046] S101, acquiring a first signal of a target power distribution line, and performing a first preprocessing on the first signal to obtain a second signal;

[0047] In an optional embodiment, the first signal of the target power distribution line may include a current value and a voltage value in a power grid signal, and may also include other relevant parameters, such as frequency, phase, etc.

[0048] In an optional embodiment, the first signal of the target distribution line can be collected by arranging a collection device, or obtained through a communication interface with a power grid monitoring center. The collection device can be a current transformer, a voltage transformer or other type of sensor for real-time monitoring of the operating status of the power grid.

[0049] In an optional embodiment, the first preprocessing step includes filtering the first signal to remove noise and interference, thereby improving the clarity and accuracy of the signal. The filtering process can adopt a variety of methods, such as low-pass filtering, high-pass filtering, band-pass filtering or band-stop filtering, etc. The specific selection depends on the characteristics of the signal and the analysis requirements.

[0050] In the embodiment of the present application, collection points are arranged at both ends of the distribution line, and a relay collection point is arranged in the middle of the distribution line. Each collection point monitors the power grid signal in real time through a collection device; the power grid signal includes voltage value and current value;

[0051] In the embodiment of the present application, a GPS synchronized clock is used to provide a unified time reference for all acquisition devices, a timestamp is added to each monitored signal to mark the time point of data acquisition, and a first preprocessing is performed on the collected power grid signal;

[0052] In the embodiment of the present application, the first preprocessing includes removing high-frequency noise interference and low-frequency drift components in the power grid signal, and performing time reversal on the power grid signal (ie, the first signal) to obtain an inverted signal;

[0053] In an optional embodiment, the high-frequency noise interference and low-frequency drift components in the power grid signal can be removed by low-pass filtering and high-pass filtering techniques respectively. The low-pass filter can remove the high-frequency noise in the signal, while the high-pass filter can remove the low-frequency drift components in the signal. Through such preprocessing, the accuracy of subsequent analysis can be ensured.

[0054] In another optional embodiment, the high-frequency noise interference and low-frequency drift components in the power grid signal can be removed by using a Butterworth low-pass filter and a polynomial fitting detrending method respectively;

[0055] It should be noted that the Butterworth low-pass filter is widely used in signal processing due to its smooth amplitude response characteristics, and can effectively remove high-frequency noise in power grid signals without introducing additional phase distortion. By setting an appropriate cutoff frequency, it can ensure that the useful components in the signal are retained while filtering out unwanted high-frequency interference.

[0056] It should be noted that the polynomial fitting detrending rule is a statistical method used to remove low-frequency drift components in signals. By selecting the appropriate polynomial order, the signal can be fitted to separate the trend part and the fluctuation part of the signal. This method is particularly suitable for processing signal drifts that change slowly over time, and can effectively extract periodic fluctuation components in power grid signals.

[0057] In the embodiment of the present application, by arranging collection points at both ends and in the middle of the distribution line and using high-precision current transformers and voltage transformers to monitor the power grid signal in real time, combined with the GPS synchronized clock to provide a unified time reference for all collection devices, the accuracy and timing consistency of signal collection are guaranteed, and the Butterworth low-pass filter is used to remove high-frequency noise interference and the polynomial fitting detrending method is used to eliminate low-frequency drift, thereby improving the clarity and stability of the signal, providing high-quality input data for subsequent analysis, and ensuring the reliability and accuracy of signal processing in complex power grid environments.

[0058] It should be noted that after the above filtering process, the inverted signal obtained will be used for subsequent fault analysis. The inverted signal not only removes noise and drift, but also, through time reversal, makes the fault features that were originally dispersed in time concentrated on the time axis, making it easier for analysts to identify and locate the specific time and location of the fault.

[0059] In an optional embodiment, the time reversal operation is to perform time reversal processing on the collected power grid signal to enhance the focusing of the signal at the fault point. Time reversal technology is a signal processing technology that uses the reciprocity principle of signal propagation to reverse the signal at the receiving end so that the path the signal passes through during the propagation process is reversed again, thereby forming a focusing effect at the original signal emission point. This method is particularly suitable for fault location in complex power grid environments because it can significantly improve the signal-to-noise ratio of the fault signal, thereby improving the accuracy of fault detection.

[0060] In another optional embodiment, an algorithm based on digital signal processing technology can be used to time-reverse the power grid signal (i.e., the first signal) to obtain an inverted signal. The algorithm first pre-processes the collected power grid signal, including steps such as filtering and amplification, to ensure the quality of the signal. Subsequently, the algorithm time-reverses the pre-processed signal, that is, arranges the time series of the signal in reverse order. This processing method allows the characteristic points of the signal on the time axis to be rearranged, so that the originally scattered fault features become concentrated in the time-reversed signal.

[0061] In another optional embodiment, the grid signal (i.e., the first signal) can be time-reversed to obtain the inverted signal by an algorithm based on analog signal processing technology. The algorithm uses an analog circuit to time-reverse the signal, avoiding quantization errors and processing delays that may be introduced in digital signal processing. First, the grid signal is preliminarily filtered by an analog filter to remove noise and interference. Then, an analog delay line is used to time-delay the signal to achieve time reversal. This method is particularly suitable for real-time fault detection systems because it can respond and process signals quickly, thereby shortening the time for fault location.

[0062] In an embodiment of the present application, time reversing the power grid signal includes time reversing the first preprocessed power grid signal and performing normalization processing;

[0063] Time reversal refers to the time reversal of the power grid signal S(t) at each time point to generate a reversal signal S r (t), the formula is:

[0064] S r (t) = S(Tt)

[0065] Among them, S r (t) represents the inversion signal at time point t, and T represents the signal duration;

[0066] In another optional embodiment, the signal duration refers to the time range covered from the start of signal acquisition to the end of the signal. In power system fault analysis, it usually refers to the time from the initial moment of the fault signal (such as the moment when the fault begins to occur) to the time when the signal decays to no longer have significant characteristics (such as approaching a steady state or noise level);

[0067] It should be noted that the inversion operation reverses the propagation direction of the fault signal, so that the energy of the fault point exhibits a focusing characteristic during the time reversal process.

[0068] It should be noted that by performing a time reversal operation on the power grid signal after the first preprocessing, the inversion signal reverses the propagation direction of the fault signal in the time domain, so that the energy of the fault point is focused during the reversal process, significantly enhancing the characteristic expression ability of the fault point signal. The normalization processing ensures the consistency of the inversion signal amplitude and avoids the interference of the abnormal signal amplitude on subsequent analysis. By accurately defining the signal duration, the key characteristic range of the fault signal is focused, which effectively reduces the influence of irrelevant noise on fault analysis and provides high-quality input signals for complex fault location and category judgment.

[0069] S102, establishing a first characteristic matrix, and obtaining a third signal according to the second signal;

[0070] In the embodiment of the present application, establishing a first characteristic matrix, and obtaining a third signal according to the second signal includes:

[0071] The first characteristic matrix is ​​an arbitrary matrix simulating the propagation characteristics of the voltage and current at the fault point;

[0072] The second signal is a signal obtained by performing a first preprocessing on the first signal collected at a plurality of collection points.

[0073] In an optional embodiment, the first characteristic matrix can be constructed using a method based on a physical model, for example, using the topological structure and component parameters of the distribution network to simulate the propagation characteristics of voltage and current in the power grid. In this way, the voltage and current changes at each point in the power grid can be more accurately reflected, thereby providing a more accurate model basis for fault analysis.

[0074] In another optional embodiment, the construction of the first characteristic matrix can also be combined with machine learning technology to optimize the matrix parameters through training data sets to adapt to different types of distribution network structures and operating conditions. This method can improve the accuracy and robustness of fault detection and location, especially in the face of complex and changeable power grid environments.

[0075] In another optional embodiment, the construction of the first characteristic matrix can also describe the propagation characteristics of the voltage and current of the fault point by constructing a BLT supermatrix, substituting the inverted signal into the BLT supermatrix to extract the test signal of each acquisition point and the fault point, and the accurate value of the voltage and current of the fault point can be calculated through the inverse matrix of the BLT supermatrix. This method utilizes the characteristics of matrix operations, and the electrical characteristics of the fault point can be effectively analyzed and determined as a whole through the inverse matrix operation, thereby providing support for the precise positioning of the fault point.

[0076] In the embodiment of the present application, a BLT supermatrix is ​​constructed to describe the propagation characteristics of the voltage and current at the fault point, and the inversion signal is substituted into the BLT supermatrix to extract the test signal of each acquisition point and the fault point;

[0077] Exemplarily, constructing a BLT supermatrix to describe the propagation characteristics of voltage and current at a fault point includes:

[0078] Establishing the physical parameter model of the transmission line includes the geometric parameters and electrical parameters of the transmission line;

[0079] Furthermore, geometric parameters refer to obtaining and measuring the total length of the distribution line through GIS, measuring the horizontal spacing between adjacent conductors and the vertical ground distance of each phase conductor according to the actual project;

[0080] It should be noted that electrical parameters refer to measuring the radius of a single conductor, calculating the resistance of the conductor per unit length based on the conductor material and cross-sectional area, calculating the inductance per unit length of each phase conductor, calculating the capacitance per unit length of each phase conductor to the ground and calculating the conductance per unit length based on the capacitance per unit length, and calculating the propagation speed of the power grid signal in the transmission line;

[0081] Furthermore, the transmission line is equivalent to a multi-section lumped parameter model, the length of each section is evenly divided according to the total length of the transmission line, the physical parameters of each section are calculated, and the parameters of each section constitute a π-type network of lumped parameters;

[0082] It should be noted that the π-type network is an equivalent form of the lumped parameter model of the transmission line, which is used to describe the electrical characteristics of each section of the line. Its basic principle is to equate the distributed parameter model to lumped parameters and represent the characteristics of a section of the transmission line in the form of a network.

[0083] Furthermore, a field impedance analyzer is used to compare the difference between the calculated value and the actual test value, and a signal generator is used to inject a test grid signal into the transmission line to measure the signal propagation time and verify the propagation speed;

[0084] Furthermore, the spatial variation rate of voltage propagation and the spatial variation rate of current propagation in the transmission line are calculated;

[0085] Furthermore, by using the transmission line wave equation of the distributed parameter model, the power grid signal is transformed from the time domain to the frequency domain, and the spatial rate of change of voltage propagation in the transmission line is calculated. and the spatial rate of change of current propagation The transmission line wave equation is:

[0086]

[0087]

[0088] Among them, V(x,t) represents the voltage value of the spatial coordinate variable at time point t, and I(x,t) represents the current value of the spatial coordinate variable at time point t. represents partial derivative, which is used to represent the rate of change of a function with respect to a variable, r, l, g, c represent the resistance, inductance, conductance and capacitance of the transmission line per unit length, ω represents the angular frequency of the power grid signal, j represents the imaginary unit, which represents the contribution of the imaginary part in the complex number;

[0089] It should be noted that x represents the spatial coordinate variable in the transmission line, which is used to describe the spatial position of a point in the transmission line relative to the collection point. When x = 0, it represents the reference position of the collection point. When x>0, it represents the spatial distance from the collection point along the positive direction of the line.

[0090] Furthermore, the spatial rate of change of voltage and current is used to determine the characteristic impedance Z in the propagation of power grid signals. c (f) and propagation constant γ(f);

[0091] It should be noted that characteristic impedance refers to the ratio between voltage and current, which describes the propagation impedance Z of power grid signals in the transmission line. c (f), the formula is:

[0092]

[0093] Where f represents the frequency of the power grid signal, Zc The unit of (f) is Ω, and the value is usually between 300Ω and 500Ω. The specific value is determined by the geometric structure and material characteristics of the circuit;

[0094] It should be noted that the propagation constant refers to the propagation constant γ(f) derived by the following formula:

[0095]

[0096] Among them, μ(f) represents the real part of the propagation constant, which corresponds to the attenuation coefficient of the grid signal amplitude. represents the imaginary part of the propagation constant, the imaginary part corresponds to the coefficient of change of the grid signal phase, and j represents the imaginary unit;

[0097] It should be noted that the introduction of propagation constant and characteristic impedance enables the formula to dynamically describe the propagation law of signals in transmission lines, especially its influence on amplitude attenuation and phase change. By introducing frequency dependence, the adaptability of the model to multi-frequency signals is enhanced, making the formula more flexible and able to adapt to the actual situation of frequency fluctuations in complex power systems.

[0098] Furthermore, a sub-matrix of the BLT supermatrix is ​​constructed based on the characteristic impedance and propagation constant of the transmission line;

[0099] It should be noted that the sub-matrices for constructing the BLT supermatrix include the voltage-current propagation matrix, the propagation matrix, the nonlinear voltage coupling matrix, and the nonlinear current propagation matrix;

[0100] The voltage-current propagation matrix A(x) represents the attenuation characteristics of voltage and current, and the formula is:

[0101]

[0102]

[0103]

[0104] Among them, e γ(f)x represents the amplitude growth and phase rotation of the signal, e -γ(f)x Indicates the signal's amplitude attenuation and phase reverse rotation;

[0105] Furthermore, by introducing the hyperbolic cosine function, the nonlinear change of the amplitude during the propagation of the power grid signal is captured, which improves the accuracy of the signal description, especially in the long-distance propagation scenario, and can more realistically reflect the propagation characteristics;

[0106] The formula of nonlinear voltage coupling matrix B(x) is:

[0107]

[0108] Among them, Z c (f) represents the propagation impedance in the transmission line;

[0109] The formula of nonlinear current propagation matrix C(x) is:

[0110]

[0111] in, Describe the coupling effect of voltage on current;

[0112] Furthermore, the BLT supermatrix is ​​constructed by combining the sub-matrices, and the cumulative effect is introduced through the frequency-dependent attenuation coefficient to describe the propagation characteristics of voltage and current. The formula is:

[0113]

[0114] Among them, V(x,f) and I(x,f) represent the voltage and current at the spatial coordinate variable and frequency f, respectively, μ(f) represents the real part of the propagation constant, which corresponds to the attenuation coefficient of the power grid signal amplitude, V(0) and I(0) represent the voltage and current at the sampling point at time t, respectively, and d represents the small increment of the spatial coordinate variable;

[0115] It should be noted that the BLT supermatrix integrates the modeling capabilities of multi-frequency dependence, nonlinear propagation characteristics and cumulative attenuation effects. The formula can comprehensively and accurately describe the propagation law of signals in complex transmission lines, including the dynamic characteristics of voltage and current and the energy loss of signals. The formula supports signal propagation analysis in complex scenarios, such as the modeling needs of long-distance, high-frequency fluctuations and nonlinear power systems;

[0116] It should be noted that by introducing the attenuation coefficient, the formula can accurately describe the energy loss of power grid signals during the propagation process. Especially in long-distance transmission lines, the existence of the attenuation coefficient improves the simulation accuracy of signal amplitude changes and significantly enhances the practical applicability of the formula. This improvement not only solves the deficiency of ignoring long-distance attenuation in the prior art, but can also be used to analyze the characteristics of signals in harsh propagation environments.

[0117] It should be noted that by constructing a BLT supermatrix to describe the voltage and current propagation characteristics of the fault point, the present invention comprehensively combines the geometric parameters and electrical parameters of the transmission line to accurately model the propagation law of the signal, especially accurately describes the spatial change rate, amplitude attenuation and phase change characteristics of voltage and current under complex topological structures, introduces characteristic impedance, propagation constant and frequency dependence, and enhances the model's adaptability to multi-frequency signals. At the same time, through the hyperbolic cosine function and nonlinear coupling matrix, the dynamic nonlinear propagation characteristics of the signal in a long-distance, high-frequency fluctuation environment are captured, and the signal description accuracy and energy attenuation simulation capability are significantly improved. This improvement effectively overcomes the problem of insufficient modeling of complex power system signals in the prior art, and provides an efficient and accurate solution for signal propagation analysis and fault location of long-distance transmission lines.

[0118] Furthermore, the inverted signal is substituted into the BLT super matrix to extract the test signal of each acquisition point and fault point including,

[0119] The inversion signal is used as the excitation signal of the voltage and current of the acquisition point and input into the BLT super matrix. The loading expression of the excitation signal is:

[0120] [V(0,t)]=V r (t)

[0121]

[0122] Among them, V r (t) represents the reverse voltage signal at time point t, Z c (f) represents the characteristic impedance of power grid signal propagation;

[0123] Furthermore, it is necessary to ensure that the timing of the excitation signal is consistent with the time step in the BLT supermatrix. If the signal sampling frequency is different from the model calculation step, a linear interpolation formula is required for interpolation adjustment;

[0124] Furthermore, according to actual needs, the safe amplitude range of the excitation signal is set to check whether the amplitude of the excitation signal meets the line safety standards, and the signal amplitude outside the safe amplitude range is eliminated to avoid nonlinear effects on the model;

[0125] Calculate the reflected signal, the formula is:

[0126]

[0127] Among them, R n (t) represents the reflected signal, V r (t) represents the reverse voltage signal at time point t, I r (t) represents the reverse current signal at time point t;

[0128] Furthermore, according to circuit theory, the reflected signal satisfies the energy conservation principle, and the total energy of the injected excitation signal propagated and reflected in the transmission line cannot exceed the total energy of the input signal;

[0129] Calculate the total energy of the reflected signal and the excitation signal. If the reflected signal is greater than the excitation signal, the user will be prompted: "Energy conservation verification failed, please check the input signal amplitude". Otherwise, perform amplitude calculation.

[0130] Furthermore, the maximum amplitude of the reflected signal and the excitation signal is calculated, and the amplitude ratio β between the reflected signal and the excitation signal is calculated. If the amplitude ratio satisfies 0≤β≤1, the reflection delay time is compared. Otherwise, the user is prompted that "the amplitude ratio of the reflected signal exceeds the physical range, please check the line impedance";

[0131] It should be noted that the length of the transmission line, the signal propagation speed and the actually measured reflection delay time are input, and the error between the reflection delay time and the theoretical delay time is compared. If the error between the reflection delay time and the theoretical delay time is greater than 5% of the theoretical value, the user is prompted "the delay time deviation is too large, please check the line parameter settings and signal sampling accuracy". Otherwise, if there is no abnormality in the signal check, the user is prompted "the signal verification is passed, and the verification is completed";

[0132] It should be noted that the theoretical delay time refers to the time required for a signal to propagate in a transmission line to the fault point and return to the starting point. It is calculated based on the physical law of electromagnetic wave propagation in a transmission line. This time depends on the length of the transmission line and the propagation speed of the signal.

[0133] Furthermore, the verified excitation signal is loaded into the BLT supermatrix, the dynamic changes of the voltage and current at the acquisition point are recorded and stored in the form of time series;

[0134] Furthermore, according to each concentrated parameter model, discrete intervals are set, and each line section is divided into multiple discrete points at equal intervals. Each discrete point is used as a fault point. The fault point represents the fault location in each line section, and all fault points are integrated into a fault point set.

[0135] Furthermore, for each fault point in the transmission line, the distance between the fault point location and the acquisition point and the electrical parameters of the transmission line are calculated through the BLT super matrix to calculate the voltage and current of the fault point;

[0136] Furthermore, starting from the collection point, the fault point position is updated point by point according to the set discrete interval, and the traversal order is linearly increasing. If the distance between the collection point and the u-th fault point in each section of the transmission line is less than or equal to the length of each section of the transmission line, the traversal continues to the next fault point position, otherwise the traversal stops;

[0137] Furthermore, based on the traversal results, the test signals of each acquisition point and fault point are extracted and stored;

[0138] It should be noted that the test signal includes voltage amplitude, voltage phase, current amplitude and current phase.

[0139] It should be noted that by loading the inversion signal as the excitation signal into the BLT supermatrix and utilizing its ability to accurately describe the voltage and current propagation characteristics, the test signals of the acquisition point and the fault point are extracted, and the signal timing and step consistency are adjusted through linear interpolation to ensure the accurate input of the excitation signal. The integrity and physical rationality of the signal are verified by using the energy conservation principle and the amplitude ratio and delay time verification mechanism, which significantly improves the reliability of signal processing. The voltage and current characteristics of all fault points are systematically calculated by adopting a point-by-point progressive traversal method combined with discretization processing, ensuring the accuracy and efficiency of fault point positioning in complex distribution networks.

[0140] S103, inputting the third signal into the pre-trained first judgment network, obtaining the fault category of the fault point, and tracing the evolution of the fault location.

[0141] In the embodiment of the present application, inputting the third signal into the pre-trained first judgment network, obtaining the fault category of the fault point, and tracing the evolution of the fault location includes:

[0142] The input of the first judgment network is the third signal, and the output is the probability distribution of the fault category of the fault point or any network that can directly or indirectly solve the relevant parameters of the probability distribution of the fault category;

[0143] Performing fault location evolution tracing includes performing fault location evolution tracing through a preset first evolution tracing strategy.

[0144] In an optional embodiment, the first judgment network can be implemented by constructing a Siamese network, which can effectively identify and distinguish different types of faults by learning the characteristics of the fault signal. The Siamese network consists of two identical sub-networks that share the same parameters and process the input signal in parallel, and the fault category is judged by comparing the output difference of the two sub-networks. This structure is particularly suitable for handling fault identification problems in distribution networks because it can capture subtle differences in fault signals.

[0145] In another optional embodiment, the first judgment network can also be implemented by constructing a convolutional neural network (CNN), which extracts the spatial features of the fault signal through a convolution operation and is suitable for processing data with spatial correlation. CNN performs well in the fields of image recognition and signal processing, and can automatically learn and extract the features of the fault signal, thereby improving the accuracy of fault classification.

[0146] In another optional embodiment, the first judgment network can also be implemented by constructing a recurrent neural network (RNN), which is particularly suitable for processing time series data. RNN can memorize previous information and use this information to influence subsequent outputs, which is very useful for analyzing fault signals that change over time. In distribution network fault analysis, fault signals often have temporal continuity, and RNN can capture this time dependency, thereby more accurately locating faults and tracing their evolution.

[0147] In the embodiment of the present application, the first judgment network constructs a Siamese network for design.

[0148] In the embodiment of the present application, the first evolution tracing strategy includes:

[0149] After determining the fault type of the fault point, simulating the current values ​​of different fault points based on the first characteristic matrix, and establishing a first fault judgment set according to the current values;

[0150] The first fault judgment set is a data set sorted according to the current energy size;

[0151] A first frequency domain energy threshold is preset, and the first fault judgment set is judged one by one in a fixed order based on the first frequency domain energy threshold.

[0152] In the embodiment of the present application, the first evolution tracing strategy also includes:

[0153] If the frequency domain energy of the fault point in the first fault judgment set exceeds the set energy threshold, it is recorded as a reasonable fault point;

[0154] Otherwise, the first fault judgment set is judged one by one in a fixed order.

[0155] In the embodiment of the present application, the first evolution tracing strategy also includes:

[0156] According to the distance between the reasonable fault point and the actual observed fault point and the signal propagation speed, the time deviation is converted into the distance deviation by using the difference between the theoretical propagation time of the signal and the actual observed time;

[0157] The coordinates of the reasonable fault point are adjusted according to the distance deviation, and the optimization target is constructed using the least squares method. The sum of the squares of the deviations between the reasonable fault point response after correction and the actual observed response is minimized. The final fault point position is obtained through iterative optimization using the gradient descent method.

[0158] Tracing the evolution of the fault location includes: updating the first characteristic matrix by taking the final fault point location as a reasonable fault point location.

[0159] Exemplarily, the fault category of the fault point determined by constructing a Siamese network based on the extracted test signal includes:

[0160] Normalize the amplitudes of all test signals, take the test signal at the acquisition point as the pre-fault signal, and take the test signal at the fault point as the fault signal;

[0161] Furthermore, historical test signals are collected, and each pair of signals in the historical test signals is assigned a label and divided into a training set and a test set;

[0162] In an optional embodiment, each pair of signals is assigned a label, that is, if both groups of signals are pre-fault signals, the label is 0, and if both groups of signals include a fault signal, the label is 1;

[0163] Furthermore, uniform distribution is used to set the initial values ​​of all parameters in the Siamese network, the Adam optimizer is used, the learning rate and momentum parameters are set, the contrast loss function is used, and the preset threshold is set. If the signal pair label is 0, the loss function encourages the Siamese network to reduce the feature distance. If the signal pair label is 1, the loss function encourages the Siamese network to increase the feature distance to the preset threshold.

[0164] Furthermore, a set of signal pairs are randomly sampled from the training set and input into the two branches of the Siamese network. The contrast loss is calculated based on the actual labels to measure the deviation between the model's prediction results and the target. The gradient of the loss to the model parameters is calculated using the back-propagation algorithm, and the weights and biases of the network are updated through the optimizer. The number of training rounds is set according to actual needs, and the above steps are repeated to traverse all signal pairs in the training set to complete a round of training.

[0165] Furthermore, after each round of training, the validation set is input into the model to evaluate the performance of the model, the average loss value on the validation set is calculated, and the classification accuracy and misclassification rate are recorded;

[0166] Furthermore, the real-time test signal is input into the Siamese network, and the pre-fault signal and the fault signal are processed respectively by two independent Transformer encoders. The embedding layer in each Transformer encoder maps the input signal from the original dimension to a high-dimensional space, generates embedded features, and adds position codes for each time point in the position encoding layer;

[0167] In an optional embodiment, multi-head self-attention is used to calculate the dependency between input signals at different time points. Each attention head generates a query matrix, a key matrix, and a value matrix by linearly transforming the input embedding signal. The multi-head attention mechanism concatenates the results of multiple groups of attention and projects them back to the embedding dimension.

[0168] In an optional embodiment, two branches of the Siamese network are used to encode the pre-fault signal and the fault signal respectively, generate two feature vectors, calculate the feature difference vectors of the pre-fault and fault signals, use the difference vector to calculate the abnormality score of the signal pair, set the abnormality threshold, if the abnormality score is greater than the abnormality threshold, the fault point is determined to be a fault signal, and use the fully connected classifier Softmax to output the probability distribution of multiple categories of fault types, select the fault category according to the highest probability output, otherwise, the fault point is determined to be a normal signal.

[0169] In an optional embodiment, by constructing a Siamese network and combining the embedded feature learning and multi-head self-attention mechanism of the Transformer encoder, the method effectively extracts the high-dimensional feature difference between the pre-fault signal and the fault signal, and accurately measures the signal feature distance through the contrast loss function, thereby achieving efficient discrimination of complex fault categories. Normalization and label assignment enhance the training stability of the model. The Adam optimizer and back-propagation mechanism further optimize the network weights, so that the model still has good generalization ability under the condition of few samples. Finally, through the combination of anomaly scores and classifiers, the fault point can be accurately identified and the probability distribution of multiple categories of fault types can be output, which significantly improves the diagnostic accuracy and robustness in complex fault scenarios.

[0170] In an optional embodiment, the fault location inference and optimization includes simulating the current values ​​of different fault points through the BLT super matrix, calculating the current energy of each fault point determined as a fault signal by using the 2 norm, selecting the fault point with the largest energy as the preliminary fault location, and selecting the remaining fault points as candidate points;

[0171] After obtaining the preliminary fault point, the signal energy focusing verification is performed. The energy concentration characteristic of the time-reversed signal at the real fault point is used to verify the fault point by calculating the sum of the squares of the frequency domain energy. The energy threshold is set. If the frequency domain energy exceeds the set energy threshold, the preliminary fault point is considered reasonable. Otherwise, the fault point with the largest energy is reselected from the alternative points.

[0172] Furthermore, combined with the distance between the fault point and the observation point and the signal propagation speed, the time deviation is converted into a distance deviation using the difference between the theoretical propagation time of the signal and the actual observation time, and the coordinates of the fault point are adjusted. The optimization target is constructed using the least squares method, and the sum of the squares of the deviations between the theoretical response of the corrected point and the actual observed response is minimized. The final fault point position is obtained through iterative optimization using the gradient descent method.

[0173] It should be noted that by simulating the current values ​​of different fault points through the BLT supermatrix, combined with the 2-norm calculation and signal energy focusing verification, the preliminary fault position with the largest energy is accurately screened and determined, effectively reducing misjudgment in complex fault scenarios, and using time-reversal signals to enhance the feature focusing of the fault point. The rationality of the preliminary fault point is ensured through frequency domain energy verification, and the positioning accuracy is further improved by combining the signal propagation time deviation correction. Finally, through the least squares method and gradient descent iterative optimization, the deviation between the theoretical response and the actual observed response is minimized, achieving high-precision inference and optimization of the fault location, and significantly improving the reliability and robustness of fault location in complex power grid environments.

[0174] In summary, the present invention proposes an evolutionary source tracing analysis method for complex faults in distribution network coefficients, obtaining a first signal of the target distribution line, and performing a first preprocessing on the first signal to obtain a second signal; establishing a first characteristic matrix, and obtaining a third signal according to the second signal; inputting the third signal into a pre-trained first judgment network, obtaining the fault category of the fault point, and performing evolutionary source tracing of the fault location. Through the time reversal operation, the focusing of the power grid fault signal at the fault point is significantly enhanced. Combined with the BLT super-matrix modeling, the amplitude attenuation, phase change and energy reflection characteristics of signal propagation in complex power grids are accurately described. Through the deep feature learning of the Siamese network, combined with energy focusing verification and least squares optimization, the fault classification and location accuracy in complex fault scenarios are improved.

[0175] Example 2

[0176] In a preferred embodiment, the design Figure 2The flowchart of signal verification shown in the figure, wherein 201 is to collect the grid signal in the distribution line in real time and preprocess it, and time-reversing the grid signal. This step involves collecting the grid signal in real time and preprocessing it, including time reversal. 202 is to construct a BLT super-matrix to describe the propagation characteristics of the voltage and current at the fault point. In this step, a BLT super-matrix is ​​constructed to describe the propagation characteristics of the voltage and current at the fault point. 203 is to use the reversed signal as the excitation signal of the voltage and current of the acquisition point and input it into the BLT super-matrix. The time-reversed signal is input into the BLT super-matrix as the excitation signal. 204 is to set a safe amplitude interval for the excitation signal and determine whether the signal amplitude is within the safe amplitude interval. A safe amplitude interval is set and whether the amplitude of the input excitation signal is within this interval. 205 is to delete the signal amplitude. If the signal amplitude is not within the safe interval, the signal amplitude is deleted. 206 is to calculate the total energy of the reflected signal and the excitation signal and determine whether the reflected signal is less than the energy of the excitation signal. The total energy of the reflected signal and the excitation signal is calculated and whether the energy of the reflected signal is less than the energy of the excitation signal. 207 is to prompt the user that "Energy conservation verification failed, please check the input signal amplitude". If the energy of the reflected signal is not less than the energy of the excitation signal, the user is prompted that the energy conservation verification failed, and it is recommended to check the amplitude of the input signal. 208 is to judge whether the amplitude ratio between the reflected signal and the excitation signal meets a condition. 209 is to prompt the user that "The amplitude ratio of the reflected signal exceeds the physical range, please check the line impedance". If the amplitude ratio does not meet the condition, the user is prompted that the amplitude ratio of the reflected signal exceeds the physical range, and it is recommended to check the line impedance. 210 is to judge whether the error between the reflection delay time and the theoretical delay time is less than 5% of the theoretical value. 211 is to prompt the user that "The delay time deviation is too large, please check the line parameter settings and signal sampling accuracy". If the error is greater than 5%, the user is prompted that the delay time deviation is too large, and it is recommended to check the line parameter settings and signal sampling accuracy. 212 is to prompt the user that "the signal verification has passed, and the verification has been completed". If all conditions are met, the user is prompted that the signal verification has passed, and the verification has been completed.

[0177] It should be noted that through the above signal verification process, it can be ensured that the excitation signal input into the BLT super matrix is ​​safe and conforms to the laws of physics, thereby ensuring the accuracy and reliability of subsequent fault analysis. The implementation of the signal verification process not only improves the efficiency of fault detection, but also reduces the possibility of false alarms and missed alarms through a strict signal verification mechanism, providing solid technical support for the stable operation of the distribution network.

[0178] Example 3

[0179] This embodiment also provides an evolution tracing and source analysis system for complex faults in a distribution network, including:

[0180] A signal acquisition module, used for acquiring a first signal of a target power distribution line, and performing a first preprocessing on the first signal to obtain a second signal;

[0181] A matrix establishment module, used to establish a first characteristic matrix and obtain a third signal according to the second signal;

[0182] The identification module is used to input the third signal into the pre-trained first judgment network, obtain the fault category of the fault point, and trace the evolution of the fault location.

[0183] The above-mentioned unit modules may be embedded in or independent of a processor in a computer device in the form of hardware, or may be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to the above-mentioned modules.

[0184] This embodiment also provides a computer device, which may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 3 As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a method for tracing the evolution of complex faults of distribution network coefficients is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0185] This embodiment further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0186] Acquire a first signal of a target power distribution line, and perform a first preprocessing on the first signal to obtain a second signal;

[0187] Establishing a first characteristic matrix, and obtaining a third signal according to the second signal;

[0188] The third signal is input into the pre-trained first judgment network to obtain the fault category of the fault point and to trace the evolution of the fault location.

[0189] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

[0190] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiments of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0191] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0192] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0193] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0194] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0195] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for tracing the evolution of complex faults in distribution network coefficients, characterized in that: include: Acquire a first signal of a target power distribution line, and perform a first preprocessing on the first signal to obtain a second signal; Establishing a first characteristic matrix, and obtaining a third signal according to the second signal; The third signal is input into the pre-trained first judgment network to obtain the fault category of the fault point and to trace the evolution of the fault location.

2. The method for tracing the evolution of complex faults in distribution network coefficients according to claim 1 is characterized in that: The establishing of the first characteristic matrix and obtaining the third signal according to the second signal comprises: The first characteristic matrix is ​​an arbitrary matrix simulating the propagation characteristics of the voltage and current at the fault point; The second signal is a signal obtained by performing a first preprocessing on the first signal collected at a plurality of collection points.

3. The method for tracing the evolution of complex faults in distribution network coefficients according to claim 2 is characterized in that: The step of inputting the third signal into a pre-trained first judgment network, obtaining a fault category of the fault point, and tracing the evolution of the fault location includes: The input of the first judgment network is the third signal, and the output is the probability distribution of the fault category of the fault point or any network that can directly or indirectly solve the relevant parameters of the probability distribution of the fault category; The performing fault location evolution tracing includes performing fault location evolution tracing by using a preset first evolution tracing strategy.

4. The method for tracing the evolution of complex faults in distribution network coefficients according to claim 3 is characterized in that: The first evolution traceability strategy includes: After determining the fault type of the fault point, simulating the current values ​​of different fault points based on the first characteristic matrix, and establishing a first fault judgment set according to the current values; The first fault judgment set is a data set sorted according to the magnitude of current energy; A first frequency domain energy threshold is preset, and the first fault judgment set is judged one by one in a fixed order based on the first frequency domain energy threshold.

5. The method for tracing the evolution of complex faults in distribution network coefficients according to claim 4 is characterized in that: The first evolution traceability strategy also includes: If the frequency domain energy of the fault point in the first fault judgment set exceeds the set energy threshold, it is recorded as a reasonable fault point; Otherwise, the first fault judgment set is judged one by one in a fixed order.

6. The method for tracing the evolution of complex faults in distribution network coefficients according to claim 5 is characterized in that: The first evolution traceability strategy also includes: According to the distance between the reasonable fault point and the actually observed fault point and the signal propagation speed, the time deviation is converted into a distance deviation by using the difference between the theoretical propagation time of the signal and the actually observed time; The coordinates of the reasonable fault point are adjusted according to the distance deviation, and the optimization target is constructed using the least squares method. The sum of the squares of the deviations between the reasonable fault point response of the corrected point and the actual observed response is minimized. The final fault point position is obtained through iterative optimization using the gradient descent method.

7. The method for tracing the evolution of complex faults in distribution network coefficients according to claim 6 is characterized in that: The tracing of the fault location evolution includes: updating the first characteristic matrix by taking the final fault point location as a reasonable fault point location.

8. An evolution tracing and analysis system for complex faults in distribution network coefficients, characterized in that: Said include: A signal acquisition module, used for acquiring a first signal of a target power distribution line, and performing a first preprocessing on the first signal to obtain a second signal; A matrix establishment module, used to establish a first characteristic matrix and obtain a third signal according to the second signal; The identification module is used to input the third signal into the pre-trained first judgment network, obtain the fault category of the fault point, and trace the evolution of the fault location.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images