Power grid fault detecting and positioning method and system for intelligent power grid system

Abstract

The invention provides a power grid fault detection and positioning method and system for an intelligent power grid system. The method comprises the following steps: collecting a fault traveling wave signal when a power distribution line has a fault; preprocessing the fault traveling wave signal to obtain a preprocessed target fault traveling wave signal; carrying out traveling wave front edge detection on the target fault traveling wave signal, and determining the signal arrival time of the fault traveling wave signal; fault signal features are extracted from the target fault traveling wave signals through a frequency domain analysis method and a time domain analysis method; estimating the fault line characteristics of the fault line with the fault in combination with the fault signal characteristics and the signal arrival time; outputting the line type and the fault type of the fault line through a feature recognition model; selecting an optimal fault positioning algorithm by using a decision tree algorithm based on the line type and the fault type; and determining the line position of the fault line according to the target fault traveling wave signal by adopting an optimal fault positioning algorithm. The method has the effect of intelligently selecting the optimal positioning algorithm according to the line type.

Description

Technical Field

[0001] The present invention belongs to the technical field of power grid fault detection, and in particular relates to a power grid fault detection and positioning method and system for a smart grid system. Background Art

[0002] With the acceleration of urbanization and the growth of electricity demand, the power grid structure has become increasingly complex. The types of lines in different regions and application scenarios are showing a diversified trend, including high-voltage transmission lines, medium- and low-voltage distribution lines, underground cables, and hybrid lines. This complexity poses a huge challenge to fault location. At the same time, power users are constantly increasing their requirements for power supply reliability, requiring faster and more accurate fault location capabilities.

[0003] Since each type of line has its own unique electrical characteristics and fault characteristics. Secondly, the same type of line will also show different characteristics under different environments and operating conditions, which further increases the difficulty of selecting a suitable algorithm. Traditional fault location methods are often designed for a specific type of line, lacking sufficient flexibility and adaptability, and are difficult to adapt to the characteristics of different line types, resulting in low positioning accuracy and low efficiency. Summary of the invention

[0004] The present invention provides a power grid fault detection and positioning method and system for a smart grid system, so as to solve the problem that traditional fault positioning methods are difficult to adapt to the characteristics of different line types, resulting in low positioning accuracy and low efficiency.

[0005] In a first aspect, the present invention provides a method for detecting and locating a power grid fault in a smart grid system, the method comprising the following steps: The fault traveling wave signal when a distribution line fails is collected by a traveling wave recording device deployed in the smart grid system; Preprocessing the fault traveling wave signal to obtain a preprocessed target fault traveling wave signal; Performing traveling wave front detection on the target fault traveling wave signal to determine the signal arrival time of the fault traveling wave signal; Extracting fault signal features from the target fault traveling wave signal by using a frequency domain analysis method and a time domain analysis method; estimating a fault line characteristic of a faulty line where a fault occurs in combination with the fault signal characteristic and the signal arrival time; Inputting the fault signal feature and the fault line characteristic into a pre-trained feature recognition model, and outputting the line type and fault type of the fault line through the feature recognition model; Selecting an optimal fault location algorithm based on the line type and the fault type using a decision tree algorithm; The line position of the fault line is determined according to the target fault traveling wave signal and using the optimal fault location algorithm.

[0006] Optionally, the preprocessing of the fault traveling wave signal to obtain a preprocessed target fault traveling wave signal comprises the following steps: Using wavelet transform method to perform noise reduction processing on the fault traveling wave signal; A bandpass filter is used to filter the fault traveling wave signal after noise reduction to obtain a target fault traveling wave signal.

[0007] Optionally, performing traveling wave front edge detection on the target fault traveling wave signal to determine the signal arrival time of the fault traveling wave signal comprises the following steps: Using a sliding window method to segment the target fault traveling wave signal into multiple traveling wave signal segments; For each of the traveling wave signal segments, calculating and obtaining the signal statistical characteristics and the first-order derivative of the traveling wave signal segment; If the signal statistical feature does not exceed a preset feature threshold and / or the signal first-order derivative does not exceed a preset derivative threshold, determining that there is no traveling wave front in the traveling wave signal segment; If the signal statistical feature exceeds the feature threshold and the first-order derivative of the signal exceeds the derivative threshold, determining that the traveling wave front exists in the traveling wave signal segment, and marking the traveling wave signal segment with the traveling wave front as a target traveling wave signal segment; The signal arrival time of the fault traveling wave signal is calculated based on the target traveling wave signal segment.

[0008] Optionally, the calculating the signal arrival time of the fault traveling wave signal based on the target traveling wave signal segment comprises the following steps: Improving the time resolution of the target traveling wave signal segment by interpolation; Calculating the signal cross-correlation function between the target traveling wave signal segment and the preset template signal according to the correlation analysis method; The time corresponding to the function peak position of the signal cross-correlation function is used as the signal arrival time of the fault traveling wave signal.

[0009] Optionally, the extracting the fault signal feature from the target fault traveling wave signal by using the frequency domain analysis method and the time domain analysis method comprises the following steps: Performing time domain analysis on the target fault traveling wave signal, and statistically analyzing the time domain characteristics of the target fault traveling wave signal, wherein the time domain characteristics include signal mean, signal variance, signal skewness, and signal kurtosis; Performing frequency domain analysis on the target fault traveling wave signal by Fourier transform, and extracting frequency domain features of the target fault traveling wave signal, wherein the frequency domain features include spectrum distribution, harmonic content and power spectrum density; The time domain features and the frequency domain features are integrated into the fault signal features of the target fault traveling wave signal.

[0010] Optionally, the feature recognition model includes a line type recognition module and a fault type recognition module, the line type recognition module is constructed based on a support vector machine, and the fault type recognition module is constructed based on a convolutional neural network.

[0011] Optionally, the selecting an optimal fault location algorithm based on the line type and the fault type and using a decision tree algorithm comprises the following steps: Use one-hot encoding to convert all basic line types and all basic fault types into numerical features; Calculate the information gain between all the numerical features; All fault location algorithms are taken as leaf nodes, the information gain is used as the splitting criterion, and the CART algorithm is used to recursively select the best numerical features from the root node for splitting until the number of node samples is less than a preset threshold, thereby constructing an algorithm selection decision tree model; The line type and the fault type are input into the algorithm selection decision tree model, and an optimal fault location algorithm is output through the algorithm selection decision tree model.

[0012] Optionally, the fault location algorithm includes a traveling wave method, an impedance method, an injection method, a traveling wave-traveling wave correlation method and a multi-terminal synchronous measurement method.

[0013] In a second aspect, the present invention also provides a power grid fault detection and positioning system for a smart grid system, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the computer program, the power grid fault detection and positioning method for a smart grid system as described in the first aspect is implemented.

[0014] In a third aspect, the present invention further provides a computer storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of the power grid fault detection and positioning method for a smart grid system described in the first aspect are implemented.

[0015] The beneficial effects of the present invention are: The present invention can automatically select the optimal fault location algorithm by intelligently analyzing the fault signal characteristics and line characteristics, greatly improving the accuracy and adaptability of fault location. This flexible method can effectively cope with various complex line structures, including overhead lines, underground cables and hybrid lines, etc., and overcomes the limitations of traditional fixed algorithms when facing diversified lines. Secondly, the present invention adopts advanced signal processing technology and artificial intelligence methods, which can extract more valuable information from fault traveling wave signals and improve the accuracy and efficiency of fault feature identification. Furthermore, the present invention introduces a decision tree algorithm to select the optimal fault location algorithm. This intelligent selection process not only improves the positioning efficiency, but also can be continuously optimized with the introduction of new lines and changes in the power grid structure, and has good scalability and adaptability. In addition, the overall process design of the present invention fully considers the actual application needs, from signal acquisition, preprocessing to final positioning, forming a complete set of intelligent fault location solutions, greatly improving the operation reliability and maintenance efficiency of the power system. In general, the present invention provides an innovative solution to the problem of fault location in a complex power grid environment, effectively solves the problem of algorithm selection difficulties caused by complex line structures, and provides a strong guarantee for the safe and stable operation of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of a flow chart of a method for detecting and locating power grid faults in a smart grid system in one embodiment of the present application. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments in the present application belong to the scope of protection of this application.

[0018] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0019] Figure 1FIG. 1 is a flow chart of a method for detecting and locating a power grid fault in a smart grid system in one embodiment. It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps in the above method may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps. Figure 1 As shown, a power grid fault detection and positioning method for a smart grid system disclosed in the present invention specifically includes the following steps: S101. Collect fault traveling wave signals when a distribution line fails through a traveling wave recording device deployed in the smart grid system.

[0020] Among them, traveling wave recording devices are deployed in smart grid systems. These devices are usually installed at key nodes of distribution lines, such as substations and branch points. When a distribution line fails, high-frequency electromagnetic waves are generated at the fault point, which propagate along the line at the speed of light. Traveling wave recording devices can capture these high-frequency signals and convert them into digital signals for storage and analysis. Traveling wave recording devices usually use analog-to-digital converters with high sampling rates (such as 10MHz or higher) to ensure that rapidly changing traveling wave signals can be accurately captured. The recorded signals usually include instantaneous values ​​of voltage and current, and sometimes also zero-sequence components. In order to improve the quality of the signal, traveling wave recording devices are usually equipped with filters to remove power frequency signals and other low-frequency interference. In addition, in order to ensure time synchronization, these devices are usually equipped with GPS clocks to achieve microsecond time accuracy. In practical applications, multiple traveling wave recording devices may be deployed on a distribution line to improve positioning accuracy through multi-point measurement. For example, on a 10-kilometer-long distribution line, a traveling wave recording device may be installed at the starting end, the middle point, and the terminal. When a fault occurs, these devices will trigger recording at the same time to capture the traveling wave signal generated by the fault. The collected raw signals are usually stored in a high-precision digital format, such as 16-bit or 24-bit sampling accuracy, to retain the detailed characteristics of the signal. These raw data will then be transmitted to the central processing system for further analysis. In this way, high-quality, high-time resolution fault traveling wave signals can be obtained, providing a reliable data basis for subsequent fault location analysis.

[0021] S102. Preprocess the fault traveling wave signal to obtain a preprocessed target fault traveling wave signal.

[0022] Among them, the main purpose of preprocessing is to remove noise and interference from the signal, improve the signal-to-noise ratio of the signal, and highlight the fault characteristics. First, data calibration is required to convert the raw data into actual physical quantities, such as voltage (V) or current (A). This usually involves applying a predetermined conversion coefficient. Next, a digital filter is applied to remove unwanted frequency components. Common filtering methods include Butterworth filters, Chebyshev filters, or finite impulse response (FIR) filters. For example, a high-pass filter can be used to remove 50Hz or 60Hz power frequency components, and the cutoff frequency may be set at around 1kHz. At the same time, a low-pass filter can also be used to remove high-frequency noise, and the cutoff frequency may be around 1MHz. After filtering, signal normalization may also be required to eliminate amplitude differences between different measurement points.

[0023] In order to further improve the signal quality, wavelet transform can be applied for signal denoising. Wavelet transform can analyze the signal in both time and frequency domains, and is particularly suitable for processing non-stationary signals. For example, the db4 wavelet can be used for 5-level decomposition, and then the soft threshold method can be applied to the high-frequency coefficients for denoising. The denoised signal is reconstructed by inverse wavelet transform to obtain a clear target fault traveling wave signal. In addition, baseline correction can be performed to remove the DC offset of the signal. This can be achieved by subtracting the mean of the signal or using polynomial fitting. For example, a cubic polynomial can be used to fit the baseline of the signal, and then the fitted baseline can be subtracted from the original signal.

[0024] The preprocessed target fault traveling wave signal should have a higher signal-to-noise ratio and clearer fault features, providing a good foundation for subsequent frontier detection and feature extraction. Through these preprocessing steps, the accuracy and reliability of fault location can be significantly improved. The effect of preprocessing can be evaluated by comparing the time domain waveform and spectrum of the original signal and the processed signal to ensure that the fault features are enhanced and the noise and interference are effectively suppressed.

[0025] S103. Perform traveling wave front detection on the target fault traveling wave signal to determine the signal arrival time of the fault traveling wave signal.

[0026] Among them, the accuracy of this step directly affects the accuracy of subsequent fault location. Traveling wave front detection usually adopts a strategy that combines multiple methods to improve the reliability of detection. First, the traditional threshold method can be used. This method sets a fixed or adaptive threshold, and when the signal amplitude exceeds this threshold, the traveling wave front is considered to be detected. However, the simple threshold method may be affected by noise. Therefore, the differential method can be used in combination to improve the accuracy of detection. The differential method calculates the first-order or second-order difference of the signal, and the traveling wave front usually appears as a sudden change in the differential signal. Another effective method is to use a matched filter. Design a filter that matches the shape of the ideal traveling wave front, and then convolve it with the actual signal. The peak of the convolution output corresponds to the position of the traveling wave front. The output of the matched filter can be expressed as:

[0027] Where x[k] is the input signal and h[n] is the impulse response of the matched filter.

[0028] Finally, to accurately determine the arrival time, interpolation techniques can be used. For example, cubic spline interpolation can be used near the detected leading edge to subdivide the sampling interval to obtain a sub-sampling accuracy arrival time estimate. Through the combined application of these methods, the traveling wave leading edge can be accurately detected in various complex noise environments to determine the arrival time of the signal. Accurate arrival time estimation is crucial for subsequent fault location, especially when using the time difference method for fault location. Typical detection accuracy can reach microseconds, which is crucial for accurately locating the fault point on the distribution line.

[0029] S104. Extract fault signal features from the target fault traveling wave signal through frequency domain analysis method and time domain analysis method.

[0030] Among them, in the frequency domain analysis, the fast Fourier transform (FFT) is mainly used to convert the time domain signal to the frequency domain. By observing the characteristics of the spectrum, the main frequency components of the fault signal can be identified. For example, a single-phase grounding fault usually produces obvious harmonic components near the fundamental frequency, while a phase-to-phase short circuit fault may have a characteristic peak in a higher frequency band. In the time domain analysis, the main focus is on the amplitude, duration, rise time and other characteristics of the signal. The statistical characteristics of the signal such as the mean, variance, skewness, and kurtosis can be calculated, which can reflect the overall distribution characteristics of the signal. At the same time, the envelope of the signal can also be extracted to analyze the attenuation characteristics of the signal. For transient faults, the duration and energy concentration of the signal can be calculated, which are helpful to distinguish different types of faults. In addition, the empirical mode decomposition (EMD) method can be used to decompose the signal into a series of intrinsic mode functions (IMFs), and then analyze the characteristics of each IMF. In the process of feature extraction, it is necessary to pay attention to the selection of features to be discriminative and stable to ensure that the characteristics of different types of faults can be accurately reflected. The extracted features usually include frequency domain features (such as the amplitude and phase of the main frequency components), time domain features (such as the peak value and RMS value of the signal), and time-frequency joint features (such as energy distribution diagram). These features constitute a multidimensional feature vector, which provides an important basis for subsequent fault type identification and location.

[0031] S105. Estimate the fault line characteristics of the faulty line where the fault occurs by combining the fault signal characteristics and the signal arrival time.

[0032] Among them, first, according to the difference in signal arrival time, the approximate location of the fault can be preliminarily determined. For example, if the difference in signal arrival time between two adjacent measurement points is small, then the fault point is likely to be located between the two measurement points. Secondly, the impedance characteristics of the fault point can be estimated by using the frequency domain characteristics of the fault signal. By analyzing the frequency spectrum relationship between the fault current and voltage, the equivalent impedance of the fault point can be calculated. This equivalent impedance information is very helpful for determining the type of fault (such as low-impedance grounding fault or high-impedance grounding fault). Thirdly, by analyzing the attenuation characteristics of the signal, the loss parameters of the line can be estimated. Specifically, the envelope of the signal can be fitted using an exponential decay model to obtain the attenuation coefficient. This attenuation coefficient is related to the resistance, inductance and capacitance parameters of the line and can be used to correct the line model. In addition, the propagation speed of the signal can also be used to estimate the distributed parameters of the line. The propagation speed of the traveling wave on the line is related to the inductance and capacitance per unit length of the line. The propagation speed can be calculated by the formula v=1 / √(LC), where L and C are the inductance and capacitance per unit length of the line, respectively. By comparing the actual measured propagation speed with the theoretical calculated value, the line parameters can be corrected. Finally, the fault severity and duration can be estimated by combining the polarity information and amplitude characteristics of the fault signal. These estimation results are not only helpful for fault location, but also provide a reference for subsequent fault isolation and recovery strategies. It should be noted that the accuracy of the estimation results is affected by many factors, such as measurement errors, nonlinear changes in line parameters, etc. Therefore, in practical applications, it is usually necessary to combine multiple methods for cross-validation to improve the reliability of the estimation.

[0033] S106. Input the fault signal features and the fault line characteristics into a pre-trained feature recognition model, and output the line type and fault type of the fault line through the feature recognition model.

[0034] Among them, the feature recognition model includes a line type recognition module and a fault type recognition module. The line type recognition module is built based on a support vector machine (SVM), and the fault type recognition module is built based on a convolutional neural network (CNN). The SVM model can take the extracted feature vector as input, map the data to a high-dimensional space through a kernel function (such as a radial basis function), and then find the optimal classification hyperplane in the high-dimensional space. The CNN model is particularly suitable for processing time-frequency image features and can automatically learn the local and global features of fault signals. In the model training stage, a large amount of known types of fault data is required for supervised learning. The training data should include samples of various line types (such as overhead lines, cables, hybrid lines, etc.) and fault types (such as single-phase grounding, two-phase short circuit, three-phase short circuit, etc.). In order to improve the generalization ability of the model, a cross-validation method is usually used to evaluate the model performance, and regularization techniques (such as L1 and L2 regularization) are used to prevent overfitting. The output of the model is usually the probability distribution of line type and fault type, and the category with the highest probability can be selected as the final result. In order to improve the reliability of identification, ensemble learning methods such as Bagging or Boosting can be used to integrate the prediction results of multiple models. In addition, expert knowledge can be introduced to correct the output of the model by setting some rules, especially for some rare but important fault types. The performance of the model can be evaluated by indicators such as accuracy, recall, and F1 score. In practical applications, the model needs to be updated and fine-tuned regularly to adapt to changes in the power grid system and emerging fault types. By accurately identifying the line type and fault type, it can provide an important basis for subsequent fault location and processing, and improve the efficiency and reliability of power grid operation and maintenance.

[0035] S107. Select an optimal fault location algorithm based on the line type and the fault type and using a decision tree algorithm.

[0036] Among them, the decision tree algorithm can automatically select the most suitable fault location algorithm according to different conditions. The construction process of the decision tree is usually based on expert knowledge and historical data analysis. The root node of the tree can be the line type, and then branch out according to factors such as fault type and line characteristics. For example, for a single-phase grounding fault on an overhead line, a location algorithm based on the traveling wave method may be selected; while for a phase-to-phase short circuit fault on a cable line, an impedance method may be selected. Each leaf node of the decision tree corresponds to a specific fault location algorithm. Commonly used fault location algorithms include the traveling wave method, the impedance method, and the injection method. The traveling wave method is suitable for rapidly changing transient faults, and the fault point is located by calculating the time difference between the fault traveling wave arriving at different measurement points. Its mathematical model can be expressed as: x=(L+v(t2-t1)) / 2, where x is the distance to the fault point, L is the total length of the line, v is the traveling wave propagation speed, and t1 and t2 are the times when the traveling wave arrives at both ends. The impedance law is applicable to steady-state faults. The fault location is determined by calculating the impedance of the fault point. The basic formula is: x=(V / I)*(l / Z), where V and I are the voltage and current at the fault point, l is the line length, and Z is the impedance per unit length of the line. The injection method locates the fault by injecting a signal of a specific frequency into the line and analyzing the reflection characteristics of the signal. It is particularly suitable for high-impedance faults. The decision tree algorithm not only considers the fault type and line characteristics, but also other factors, such as the type of measurement data available, environmental conditions (such as weather), historical fault statistics, etc.

[0037] S108. Determine the line position of the faulty line according to the target fault traveling wave signal and using an optimal fault location algorithm.

[0038] Among them, the location of the fault point is accurately calculated using the selected optimal algorithm.

[0039] The following are the detailed implementation methods of several commonly used fault location algorithms: 1. Traveling wave method: This method locates the fault based on the propagation time difference of the traveling wave generated by the fault on the line. Assuming there are two measurement points A and B, the total length of the line is L, the propagation speed of the traveling wave is v, and the distance from the fault point to point A is x, then:

[0040] Among them, tA and tB are the times when the fault traveling wave reaches point A and point B respectively.

[0041] To improve accuracy, data from multiple measurement points can be used and optimized using the least squares method:

[0042] Among them, xij is the fault distance calculated based on the measurement points i and j, and \(\hat{x}_{ij}\) is the estimated distance after optimization.

[0043] 2. Impedance method: The impedance method uses the impedance of the fault point to calculate the fault distance. For a single-phase grounding fault, the fault distance can be calculated using the following formula:

[0044] Where VF is the fault point voltage, IF is the fault current, Z1 is the positive sequence impedance, and L is the total line length.

[0045] To improve accuracy, the effect of load current can be taken into account and an iterative algorithm can be used:

[0046] Where IL is the load current and xn is the fault distance estimate for the nth iteration.

[0047] 3. Injection method: The injection method locates the fault by injecting a signal of a specific frequency into the line and analyzing the reflected signal. The fault distance can be calculated using the following formula:

[0048] Where v is the speed of the signal in the line, f is the frequency of the injected signal, and φ is the phase difference (in degrees) between the incident and reflected waves.

[0049] 4. Traveling wave-traveling wave correlation method: This method uses the correlation of traveling wave signals to improve positioning accuracy. First, calculate the cross-correlation function of the traveling wave signals measured at both ends:

[0050] Where sA(t) and sB(t) are the signals measured at end A and end B respectively. The fault distance can be calculated by the peak position τmax of the cross-correlation function:

[0051] 5. Multi-terminal synchronous measurement method: This method uses the synchronized data of multiple measurement points to improve positioning accuracy. Assuming there are n measurement points, the following set of equations can be established:

[0052] Among them, (x, y) is the coordinate of the fault point, (xi, yi) is the coordinate of the ith measurement point, and ti is the time when the fault wave reaches the ith measurement point. This system of equations can be solved by nonlinear least squares method.

[0053] This comprehensive, adaptive fault location method, combined with advanced signal processing technology and statistical analysis, can greatly improve the accuracy and reliability of fault location. This can not only help quickly find the fault point and reduce power outage time, but also provide accurate data support for preventive maintenance of the power grid, thereby improving the reliability and efficiency of the entire distribution system.

[0054] In one embodiment, preprocessing the fault traveling wave signal to obtain the preprocessed target fault traveling wave signal includes the following steps: The wavelet transform method is used to reduce the noise of the fault traveling wave signal; A bandpass filter is used to filter the fault traveling wave signal after noise reduction to obtain the target fault traveling wave signal.

[0055] In this embodiment, the wavelet transform has good time-frequency localization characteristics, can analyze signals at different scales, and is particularly suitable for processing non-stationary signals such as fault traveling waves. In the implementation process, first select an appropriate wavelet basis function, commonly used are Daubechies wavelet, Symlet wavelet, etc., and the signal characteristics and computational complexity must be considered when selecting. Then the original signal is decomposed at multiple scales, usually to 3-5 layers. After decomposition, wavelet coefficients of different frequency bands are obtained, including low-frequency approximate coefficients and high-frequency detail coefficients. Next, these coefficients are thresholded to remove the influence of noise. The choice of threshold is crucial, and hard threshold method or soft threshold method can be used. The mathematical expression of the hard threshold method is:

[0056] in is the wavelet coefficient, is the threshold. The expression of soft threshold method is:

[0057] The threshold can be determined using the general threshold ,in is the standard deviation of noise, and N is the signal length. After processing, the inverse wavelet transform is used to reconstruct the signal to obtain the denoised fault traveling wave signal. This method can effectively remove high-frequency noise while retaining the important features of the signal, especially the mutation characteristics at the moment of the fault. Compared with traditional filtering methods, wavelet denoising can better maintain the edge characteristics of the signal, reduce signal distortion, and improve the accuracy of subsequent analysis. In practical applications, the denoising effect can be optimized by adjusting the wavelet basis function, the number of decomposition layers, and the threshold selection strategy to adapt to different types of fault traveling wave signals.

[0058] The bandpass filter allows signals within a specific frequency range to pass through while suppressing frequency components outside this range, which is crucial for extracting the characteristic frequency band of the fault traveling wave. When designing a bandpass filter, it is first necessary to determine the main frequency range of the fault traveling wave signal, which is usually between several kilohertz and several hundred kilohertz. The type of filter can be selected from Butterworth, Chebyshev or Elliptic filters, among which Butterworth filters are often used due to their flat passband response. The order of the filter is usually selected from 4 to 8 to strike a balance between filtering effect and computational complexity. The transfer function of the bandpass filter can be expressed as:

[0059] Where K is the gain constant, and are the lower cutoff frequency and the upper cutoff frequency respectively. In digital implementation, the s domain can be converted to the z domain using a bilinear transform to obtain the difference equation of the digital filter. The filtering process can be implemented in the time domain through convolution or in the frequency domain through fast Fourier transform (FFT). The expression for time domain implementation is:

[0060] in and is the filter coefficient, M and N are the filter orders. The frequency domain implementation is to transform the signal to the frequency domain through FFT, multiply it with the filter frequency response, and then convert it back to the time domain through inverse FFT. Bandpass filtering can not only further remove residual noise, but also highlight the characteristic frequency band of the fault traveling wave and enhance the signal-to-noise ratio of the signal. This significantly improves the accuracy of subsequent fault feature extraction and location algorithms. In practical applications, the parameters of the bandpass filter can be dynamically adjusted according to different types of faults and line characteristics to obtain the best filtering effect.

[0061] In one embodiment, performing traveling wave front detection on a target fault traveling wave signal and determining the signal arrival time of the fault traveling wave signal comprises the following steps: The sliding window method is used to segment the target fault traveling wave signal into multiple traveling wave signal segments; For each traveling wave signal segment, the signal statistical characteristics and the first-order derivative of the traveling wave signal segment are calculated; If the signal statistical feature does not exceed a preset feature threshold and / or the signal first-order derivative does not exceed a preset derivative threshold, it is determined that there is no traveling wave front in the traveling wave signal segment; If the signal statistical feature exceeds the pre-feature threshold and the first-order derivative of the signal exceeds the derivative threshold, it is determined that there is a traveling wave front in the traveling wave signal segment, and the traveling wave signal segment with the traveling wave front is marked as a target traveling wave signal segment; The signal arrival time of the fault traveling wave signal is calculated based on the target traveling wave signal segment.

[0062] In this embodiment, a sliding window method is used to segment the target fault traveling wave signal into multiple traveling wave signal segments. Specifically, a fixed-size window is slid on the time axis to divide the continuous signal into a series of overlapping short-time segments. The selection of window size is usually based on the characteristic frequency and expected duration of the fault traveling wave, and the typical value may be between a few milliseconds and tens of milliseconds. The sliding step size determines the degree of overlap between adjacent windows. A smaller step size can provide more detailed analysis, but it will increase the amount of calculation. In actual operation, window functions such as rectangular windows, Hanning windows, or Hamming windows can be used to smooth signal edges and reduce spectral leakage. For a signal x(n) of length N and a window function w(m) of length M, the kth signal segment can be expressed as: , where S is the sliding step size, and m=0,1,...,M-1. This segmented processing method can capture the local characteristic changes of the signal, and is particularly suitable for analyzing non-stationary signals such as fault traveling waves, which helps to accurately locate the front of the traveling wave.

[0063] For each traveling wave signal segment, the signal statistical characteristics and the first-order derivative of the traveling wave signal segment are calculated. Signal statistical characteristics usually include mean, variance, skewness and kurtosis. The mean reflects the overall level of the signal, the variance describes the degree of fluctuation of the signal, the skewness indicates the asymmetry of the distribution, and the kurtosis reflects the degree of sharpness of the distribution. The first-order derivative of the signal reflects the rate of change of the signal and can be calculated approximately by difference: These features and derivatives can effectively capture the mutation characteristics of the traveling wave front and provide a basis for subsequent judgment.

[0064] If the statistical characteristics of the signal do not exceed the preset characteristic threshold and / or the first-order derivative of the signal does not exceed the preset derivative threshold, it is determined that there is no traveling wave front in the traveling wave signal segment. This step determines whether there is a significant signal change by setting an appropriate threshold. The characteristic threshold is usually set based on the statistical characteristics of the signal under normal operating conditions, and multiple standard deviations can be used as a reference. The derivative threshold reflects the upper limit of the signal change rate and can be determined by analyzing historical fault data. If the calculated characteristic value and derivative do not exceed the corresponding threshold, it is considered that the signal segment is in a stable state and does not contain a traveling wave front. This method can effectively filter out normal fluctuations, reduce misjudgments, and improve the accuracy of traveling wave front detection. The selection of the threshold requires a trade-off between sensitivity and false alarm rate, which can be optimized and adjusted through experimental data.

[0065] If the signal statistical feature exceeds the feature threshold and the first-order derivative of the signal exceeds the derivative threshold, it is determined that there is a traveling wave front in the traveling wave signal segment, and the traveling wave signal segment with the traveling wave front is marked as the target traveling wave signal segment. This step combines multi-dimensional judgment criteria to improve the reliability of detection. When the signal feature and derivative exceed the threshold at the same time, it indicates that the signal has not only changed significantly in amplitude, but also the rate of change exceeds the normal range, which is a typical feature of the traveling wave front. Signal segments that meet the conditions are marked as target traveling wave signal segments. These segments contain key information of the fault traveling wave, which is crucial for subsequent fault location.

[0066] The signal arrival time is usually defined as the starting point of the traveling wave front, which can be determined by a variety of methods. One common method is to find the first sampling point in the target traveling wave signal segment where the signal amplitude or energy exceeds a certain threshold. Another method is to use correlation analysis to cross-correlate the target signal segment with the preset ideal traveling wave front template. The moment when the correlation coefficient reaches the maximum value is the arrival time. After determining the arrival time, an accurate timestamp can be obtained in combination with the sampling rate. This timestamp is crucial for subsequent fault location calculations, especially when using the traveling wave method for fault location. Accurate arrival time can significantly improve the accuracy of fault location, reduce location errors, and speed up the fault recovery process.

[0067] In one embodiment, calculating the signal arrival time of the fault traveling wave signal based on the target traveling wave signal segment includes the following steps: Improve the time resolution of the target traveling wave signal segment by interpolation; Calculate the signal cross-correlation function between the target traveling wave signal segment and the preset template signal according to the correlation analysis method; The time corresponding to the function peak position of the signal cross-correlation function is taken as the signal arrival time of the fault traveling wave signal.

[0068] In this embodiment, improving the time resolution of the target traveling wave signal segment by interpolation is an effective technique for improving signal accuracy. This method inserts new data points between the original sampling points, thereby increasing the sampling rate of the signal and improving the time resolution. Common interpolation methods include linear interpolation, polynomial interpolation, and spline interpolation. Linear interpolation is the simplest method, assuming that the signal change between adjacent sampling points is linear. For time point t at and The signal value y(t) between can be calculated by the following formula:

[0069] Higher-order interpolation methods such as cubic spline interpolation can provide smoother results, but the computational complexity will also increase. The interpolation factor is usually selected from 2 to 10 times, depending on the original sampling rate and the required accuracy. This method can significantly improve the time resolution of the signal, help to more accurately locate the leading edge of the traveling wave, and thus improve the accuracy of fault location.

[0070] The cross-correlation function reflects the change in the similarity between two signals over time. For discrete signals x(n) and y(n), the cross-correlation function is defined as: , where m is the time delay. In actual calculations, fast Fourier transform (FFT) can be used to improve efficiency. The preset template signal is usually constructed based on the ideal traveling wave front characteristics, which can be a step function, a ramp function, or a more complex waveform. During the calculation process, attention should be paid to the matching of signal lengths, and zero padding may be required for shorter signals. The calculation result of the cross-correlation function reflects the change in the matching degree between the target signal and the template signal with the time offset, which provides a basis for accurately locating the arrival time of the traveling wave.

[0071] The peak of the cross-correlation function indicates the position where the target signal is most similar to the template signal, and this position is likely to be the starting point of the traveling wave front. In order to improve the accuracy, parabola fitting or other interpolation methods can be performed near the peak to obtain the peak position with sub-sampling accuracy. For example, a more accurate peak position can be obtained using the following formula using parabola fitting. After obtaining the peak position, the precise arrival time can be calculated by combining the sampling rate of the signal and the previous interpolation multiple. This method can not only accurately locate the traveling wave front, but also resist the influence of noise and interference to a certain extent, and improve the reliability of fault location. Through this precise time determination, the accuracy of subsequent fault location calculations can be significantly improved, providing important support for rapid fault diagnosis and processing of power systems.

[0072] In one embodiment, extracting fault signal features from target fault traveling wave signals by using frequency domain analysis method and time domain analysis method includes the following steps: Perform time domain analysis on the target fault traveling wave signal and count the time domain characteristics of the target fault traveling wave signal. The time domain characteristics include signal mean, signal variance, signal skewness and signal kurtosis. The target fault traveling wave signal is analyzed in the frequency domain by Fourier transform, and the frequency domain characteristics of the target fault traveling wave signal are extracted. The frequency domain characteristics include spectrum distribution, harmonic content and power spectrum density. The time domain features and frequency domain features are integrated into the fault signal features of the target fault traveling wave signal.

[0073] In this embodiment, time domain analysis of the target fault traveling wave signal and statistics of the time domain characteristics of the target fault traveling wave signal are the basic steps of fault diagnosis. The time domain characteristics directly reflect the changing characteristics of the signal in the time dimension. The signal mean μ reflects the overall level of the signal, and the calculation formula is , where N is the number of sampling points and x_i is the value of the i-th sampling point. The signal variance σ^2 describes the degree of fluctuation of the signal around the mean, and the calculation formula is The signal skewness S reflects the asymmetry of the signal distribution. Positive skewness means that the right tail of the distribution is longer, while negative skewness means the opposite. The calculation formula is: The signal kurtosis K describes the sharpness of the signal distribution. The kurtosis of the normal distribution is 3. A value greater than 3 indicates a sharper distribution, and a value less than 3 indicates a flatter distribution. The calculation formula is: These statistical features comprehensively reflect the overall shape and variation characteristics of the fault traveling wave signal, which helps to distinguish different types of faults. In practical applications, the sliding window technology can be used to calculate the time-varying values ​​of these features to capture the dynamic changes of signal characteristics. The advantages of time domain analysis are intuitiveness, simple calculation, and the ability to quickly provide the basic characteristics of the fault signal, providing an important basis for subsequent fault type judgment and location.

[0074] Performing frequency domain analysis on the target fault traveling wave signal through Fourier transform and extracting the frequency domain characteristics of the target fault traveling wave signal is a key step to deeply understand the nature of the fault signal. Fourier transform converts the time domain signal to the frequency domain and reveals the frequency composition of the signal. For discrete signals, discrete Fourier transform (DFT) is used: , where n=0,1,...,N-1, k=0,1,...,N-1. In actual calculations, the Fast Fourier Transform (FFT) algorithm is usually used to improve efficiency. The spectrum distribution reflects the distribution of signal energy on each frequency component, which can be obtained by calculating the amplitude spectrum |X(k)|. Harmonic content analysis involves calculating the amplitude of the frequency components that are integer multiples of the fundamental frequency, which can be used to evaluate the degree of distortion of the signal. The power spectral density (PSD) describes how the signal power is distributed on the frequency and can be estimated by the periodogram method: During the analysis process, it is necessary to select an appropriate window function (such as Hanning window) to reduce spectrum leakage, and consider zero padding to improve frequency resolution. Frequency domain analysis can reveal periodic components and specific frequency features in the fault signal that may be ignored by time domain analysis, which is of great significance for identifying different types of faults (such as single-phase grounding, phase-to-phase short circuit, etc.). In addition, frequency domain features are also very effective in detecting high-frequency transient and harmonic interference.

[0075] Integrate the time domain features and frequency domain features into the fault signal features of the target fault traveling wave signal. This step fuses the information in the time domain and frequency domain to form a multidimensional feature vector, which provides comprehensive data support for subsequent fault diagnosis and classification. During the integration process, the scale and importance of different features need to be considered. Feature standardization methods such as z-score standardization can be used. This can eliminate the dimensional differences between different features. Another method is to use dimensionality reduction techniques such as principal component analysis (PCA) or independent component analysis (ICA) to extract the most representative feature combination. For example, PCA can find the main direction of change of the data by solving the eigenvalues ​​and eigenvectors of the covariance matrix. The integrated eigenvector can be expressed as , where f_ti represents the time domain feature and f_fi represents the frequency domain feature. This comprehensive feature not only contains the statistical characteristics and frequency information of the signal, but also retains the potential correlation between the time domain and frequency domain features. Through this integration, the characteristics of the fault signal can be more comprehensively described, the accuracy and reliability of fault diagnosis can be improved, and strong data support can be provided for subsequent fault type identification, location and severity assessment.

[0076] In one embodiment, selecting the optimal fault location algorithm based on the line type and the fault type using a decision tree algorithm comprises the following steps: Use one-hot encoding to convert all basic line types and all basic fault types into numerical features; Calculate the information gain between all numerical features; All fault location algorithms are regarded as leaf nodes, and information gain is used as the splitting criterion. The CART algorithm is used to recursively select the best numerical features from the root node for splitting until the number of node samples is less than the preset threshold, thus constructing an algorithm selection decision tree model. The line type and fault type are input into the algorithm selection decision tree model, and the optimal fault location algorithm is output through the algorithm selection decision tree model.

[0077] In this embodiment, using one-hot encoding to convert all basic line types and all basic fault types into numerical features is a method of converting categorical variables into numerical inputs that can be directly used by machine learning algorithms. In this process, each categorical variable is converted into a binary vector whose length is equal to the number of categories, in which only one element is 1 (indicating the category) and the rest are 0. For example, if the line types include "overhead line", "cable" and "hybrid line", they can be encoded as [1,0,0], [0,1,0] and [0,0,1] respectively. Similarly, for fault types such as single-phase grounding, two-phase short circuit, three-phase short circuit, etc., they can also be encoded in a similar way. The advantage of this encoding method is that it retains the equidistance between categories and does not introduce any implicit order relationship, so that each category is treated fairly. However, one-hot encoding may also cause the dimension of the feature space to increase sharply, especially when the number of categories is large. In practical applications, it may be necessary to consider using dimensionality reduction techniques or feature selection methods to process high-dimensional features.

[0078] Calculating the information gain between all numerical features is a key step in selecting the best split feature during decision tree construction. Information gain is based on the concept of entropy in information theory and is used to measure the contribution of a feature to reducing the uncertainty of the system. For a feature A, its information gain can be expressed as IG(Y,A)=H(Y)-H(Y|A), where H(Y) is the entropy of the target variable Y, and H(Y|A) is the conditional entropy of Y when feature A is known. The entropy calculation formula is H(X)=-Σp(x)log 2 p(x), where p(x) is the probability of event x. In actual calculations, for each feature, it is necessary to traverse all its possible values, calculate the reduction in system entropy when the feature takes each value, and then take the average as the information gain of the feature. For continuous features, it is usually necessary to discretize them first, such as equal-frequency binning or equal-width binning. The process of calculating information gain may be time-consuming, especially when there are a large number of features. You can consider using parallel computing technology to speed up this process. The calculation result of information gain directly affects the structure of the decision tree. A higher information gain means that the feature is more valuable for classification and should be preferred as a split node.

[0079] Taking all fault location algorithms as leaf nodes, using information gain as the splitting criterion, and using the CART algorithm to recursively select the best numerical features for splitting from the root node until the number of node samples is less than the preset threshold, constructing an algorithm selection decision tree model is a process of automatically selecting the optimal fault location algorithm. The CART algorithm is a binary tree algorithm that selects a feature and a threshold for each split to divide the data set into two parts. At each node, the algorithm calculates the information gain of all possible splitting schemes and selects the feature and split point with the largest information gain. This process is recursive until the stopping condition is reached, such as the number of node samples is less than the preset threshold or the maximum tree depth is reached. In this application, leaf nodes represent different fault location algorithms (such as the traveling wave method, the impedance method, etc.), rather than traditional category labels. The tree growth process is actually to select the most suitable algorithm under different conditions. In order to prevent overfitting, pruning operations may be required, such as using a validation set to evaluate the performance of each subtree and selecting the subtree with the strongest generalization ability.

[0080] Inputting the line type and fault type into the algorithm selection decision tree model and outputting the optimal fault location algorithm through the algorithm selection decision tree model is the last step of the entire fault location process and is also the key link in applying the theoretical model to practical problems. In this step, the actual line type and fault type need to be converted into numerical features through the previously established unique hot encoding method. Then, these features are input into the trained decision tree model. The decision tree model will start from the root node according to these input features and pass them down layer by layer according to the splitting rules of each internal node until it reaches a leaf node. This leaf node corresponds to the optimal fault location algorithm for the current input conditions. For example, for a single-phase grounding fault of a specific overhead line, the model may select the traveling wave method as the optimal algorithm; while for a two-phase short-circuit fault of a cable line, the model may select the impedance method. This automated algorithm selection process can not only make decisions quickly, but also flexibly adjust according to actual conditions to adapt to different fault scenarios. At the same time, due to the characteristics of the decision tree model, it is easy to trace the decision path and understand why a specific algorithm is selected, which is very helpful for the interpretability of the results and the maintainability of the system.

[0081] The present invention also discloses a power grid fault detection and positioning system for a smart grid system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements a power grid fault detection and positioning method for a smart grid system as described in any one of the above-mentioned embodiments.

[0082] Among them, the processor can adopt a central processing unit (CPU). Of course, according to actual usage, other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. can also be adopted. The general-purpose processor can adopt a microprocessor or any conventional processor, etc., and this application does not impose any restrictions on this.

[0083] Among them, the memory can be an internal storage unit of a computer device, such as a hard disk or memory of a computer device, or an external storage device of a computer device, such as a plug-in hard disk, a smart memory card (SMC), a secure digital card (SD) or a flash memory card (FC) equipped on the computer device, and the memory can also be a combination of an internal storage unit and an external storage device of a computer device. The memory is used to store computer programs and other programs and data required by the computer device. The memory can also be used to temporarily store data that has been output or is to be output, and this application does not impose any restrictions on this.

[0084] The present invention also discloses a computer-readable storage medium having instructions stored thereon, which, when executed by a processor, configures the processor to execute the power grid fault detection and positioning method for a smart grid system described in any one of the above-mentioned embodiments.

[0085] Among them, the computer program can be stored in a machine-readable medium, the computer program includes computer program code, the computer program code can be in the form of source code, object code, executable file or certain middleware, etc. The machine-readable medium includes any entity or device that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the machine-readable medium includes but is not limited to the above-mentioned components.

[0086] Among them, through this computer-readable storage medium, the comprehensive fault detection method for transmission lines in the above embodiment is stored in a computer-readable storage medium, and is loaded and executed on a processor to facilitate the storage and application of the above method.

[0087] A person skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of protection of the present application is limited to these examples. In line with the concept of the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of different aspects of one or more embodiments of the present application as above, which are not provided in detail for the sake of simplicity.

[0088] One or more embodiments of the present application are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the present application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of one or more embodiments of the present application should be included in the protection scope of the present application.

Claims

1. A method for detecting and locating power grid faults in a smart grid system, characterized in that: The method comprises the following steps: The fault traveling wave signal when a distribution line fails is collected by a traveling wave recording device deployed in the smart grid system; Preprocessing the fault traveling wave signal to obtain a preprocessed target fault traveling wave signal; Performing traveling wave front detection on the target fault traveling wave signal to determine the signal arrival time of the fault traveling wave signal; Extracting fault signal features from the target fault traveling wave signal by using a frequency domain analysis method and a time domain analysis method; estimating a fault line characteristic of a faulty line where a fault occurs by combining the fault signal characteristic and the signal arrival time; Inputting the fault signal feature and the fault line characteristic into a pre-trained feature recognition model, and outputting the line type and fault type of the fault line through the feature recognition model; Selecting an optimal fault location algorithm based on the line type and the fault type using a decision tree algorithm; The line position of the fault line is determined according to the target fault traveling wave signal and using the optimal fault location algorithm.

2. The method for detecting and locating power grid faults in a smart grid system according to claim 1, characterized in that: The preprocessing of the fault traveling wave signal to obtain the preprocessed target fault traveling wave signal comprises the following steps: Using wavelet transform method to perform noise reduction processing on the fault traveling wave signal; A bandpass filter is used to filter the fault traveling wave signal after noise reduction to obtain a target fault traveling wave signal.

3. The method for detecting and locating power grid faults for a smart grid system according to claim 1, characterized in that: The performing traveling wave front edge detection on the target fault traveling wave signal to determine the signal arrival time of the fault traveling wave signal comprises the following steps: Using a sliding window method to segment the target fault traveling wave signal into multiple traveling wave signal segments; For each of the traveling wave signal segments, calculating and obtaining the signal statistical characteristics and the first-order derivative of the traveling wave signal segment; If the signal statistical feature does not exceed a preset feature threshold and / or the signal first-order derivative does not exceed a preset derivative threshold, determining that there is no traveling wave front in the traveling wave signal segment; If the signal statistical feature exceeds the feature threshold and the first-order derivative of the signal exceeds the derivative threshold, determining that the traveling wave front exists in the traveling wave signal segment, and marking the traveling wave signal segment with the traveling wave front as a target traveling wave signal segment; The signal arrival time of the fault traveling wave signal is calculated based on the target traveling wave signal segment.

4. The method for detecting and locating power grid faults for a smart grid system according to claim 3, characterized in that: The step of calculating the signal arrival time of the fault traveling wave signal based on the target traveling wave signal segment comprises the following steps: Improving the time resolution of the target traveling wave signal segment by interpolation; Calculating the signal cross-correlation function between the target traveling wave signal segment and the preset template signal according to the correlation analysis method; The time corresponding to the function peak position of the signal cross-correlation function is used as the signal arrival time of the fault traveling wave signal.

5. The method for detecting and locating power grid faults for a smart grid system according to claim 1, characterized in that: The method of extracting fault signal features from the target fault traveling wave signal by using a frequency domain analysis method and a time domain analysis method comprises the following steps: Performing time domain analysis on the target fault traveling wave signal, and statistically analyzing the time domain characteristics of the target fault traveling wave signal, wherein the time domain characteristics include signal mean, signal variance, signal skewness, and signal kurtosis; Performing frequency domain analysis on the target fault traveling wave signal by Fourier transform, and extracting frequency domain features of the target fault traveling wave signal, wherein the frequency domain features include spectrum distribution, harmonic content and power spectrum density; The time domain features and the frequency domain features are integrated into the fault signal features of the target fault traveling wave signal.

6. The method for detecting and locating power grid faults for a smart grid system according to claim 1, characterized in that: The feature recognition model includes a line type recognition module and a fault type recognition module. The line type recognition module is constructed based on a support vector machine, and the fault type recognition module is constructed based on a convolutional neural network.

7. The method for detecting and locating power grid faults for a smart grid system according to claim 1, characterized in that: The selecting of the optimal fault location algorithm based on the line type and the fault type and using a decision tree algorithm comprises the following steps: Use one-hot encoding to convert all basic line types and all basic fault types into numerical features; Calculate the information gain between all the numerical features; All fault location algorithms are taken as leaf nodes, the information gain is used as the splitting criterion, and the CART algorithm is used to recursively select the best numerical features from the root node for splitting until the number of node samples is less than a preset threshold, thereby constructing an algorithm selection decision tree model; The line type and the fault type are input into the algorithm selection decision tree model, and an optimal fault location algorithm is output through the algorithm selection decision tree model.

8. The method for detecting and locating power grid faults for a smart grid system according to claim 7, characterized in that: The fault location algorithm includes a traveling wave method, an impedance method, an injection method, a traveling wave-traveling wave correlation method and a multi-terminal synchronous measurement method.

9. A power grid fault detection and positioning system for a smart grid system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the power grid fault detection and positioning method for a smart grid system as described in any one of claims 1 to 8 is implemented.

10. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the power grid fault detection and positioning method for a smart grid system according to any one of claims 1 to 8 are implemented.

Images