A method and system for fault diagnosis of a substation communication network

Abstract

The application relates to the technical field of power system communication, and discloses a substation communication network fault diagnosis method and system. The method comprises the following steps: acquiring multi-frequency signal data, performing fast Fourier transform to obtain frequency spectrum distribution data; performing frequency band division according to the frequency spectrum distribution data to obtain an initial power ratio; performing adaptive filtering according to the initial power ratio and the multi-frequency signal data to obtain a fundamental wave amplitude fluctuation feature; performing wavelet decomposition according to the fundamental wave amplitude fluctuation feature to obtain abnormal signal time-frequency energy distribution; performing K-means clustering division according to the abnormal signal time-frequency energy distribution, matching a harmonic fault database, and obtaining a fault type identifier; performing ring network topology propagation time delay analysis according to the fault type identifier and the abnormal signal time-frequency energy distribution, positioning a fault position coordinate, and generating a fault diagnosis report. The method can accurately diagnose network faults in the face of multi-frequency complex signal interference.

Description

Technical Field

[0001] The present invention relates to the technical field of power system communications, and in particular to a fault diagnosis method and system for a substation communication network. Background Art

[0002] Currently, substation communication networks, as the core pillar of stable power system operation, bear the important responsibility of data transmission and control command delivery. Their reliability directly affects the security, stability, and efficient dispatching capabilities of the power grid. In complex electromagnetic environments, communication networks must ensure high-speed and stable data transmission to support real-time monitoring and dispatching of power systems. However, any minor communication failure can lead to serious power dispatch errors and even large-scale power outages, resulting in significant economic losses and social impacts. Therefore, developing an efficient and accurate fault diagnosis method is crucial not only to improving the reliability of power grid operation, but also to effectively reduce maintenance costs and fault recovery time.

[0003] In one existing technology, fault diagnosis for substation communication networks relies on single-signal acquisition or fixed-model analysis methods. This approach collects signal data from the substation communication network and combines it with a pre-set static mathematical model to determine faults. For example, technicians typically collect voltage or current signals from the network and extract the fundamental component or specific harmonic components through simple frequency domain analysis. These components are then compared to predefined fault thresholds to identify potential anomalies. However, in actual substation operating environments, communication network signals are often interfered with by complex signals from multiple frequency bands, including power frequency signals, low-order harmonics, high-order harmonics, and transient noise introduced by switching devices. The superposition of these signals from different frequency bands makes it difficult to accurately extract and distinguish the power ratio between the primary and interfering frequency bands. Furthermore, existing technologies exhibit poor adaptability to dynamically changing electromagnetic environments. In particular, when faced with complex interference such as high-frequency noise or transient surges, they struggle to effectively separate the primary and interfering frequency bands, making it difficult to accurately capture the subtle characteristics of abnormal signals. This limitation makes it easy for the diagnostic system to make misjudgments or missed judgments when analyzing complex signals, reducing the accuracy and reliability of fault diagnosis.

[0004] In summary, the existing technology has the problem that it is difficult to accurately diagnose network faults in the face of complex multi-band signal interference. Summary of the Invention

[0005] The present invention provides a fault diagnosis method and system for a substation communication network, so as to solve the problem that it is difficult to accurately diagnose network faults in the face of multi-band complex signal interference.

[0006] In a first aspect, in order to solve the above technical problems, the present invention provides a fault diagnosis method for a substation communication network, comprising:

[0007] Obtaining multi-band signal data in a substation communication network, performing fast Fourier transform to obtain spectrum distribution data;

[0008] According to the spectrum distribution data, calculating power spectral density ratio, performing frequency band division to obtain initial power ratio of main frequency band and interference frequency band;

[0009] According to the initial power ratio and the multi-band signal data, performing adaptive filtering to obtain fundamental amplitude fluctuation characteristics;

[0010] According to the fundamental amplitude fluctuation characteristics, performing wavelet decomposition to obtain abnormal signal time-frequency energy distribution;

[0011] According to the abnormal signal time-frequency energy distribution, performing K-means clustering division, matching a pre-established harmonic fault database to obtain a fault type identifier;

[0012] According to the fault type identifier and the abnormal signal time-frequency energy distribution, performing ring network topology propagation delay analysis to locate a position coordinate of fault occurrence, and generating a fault diagnosis report.

[0013] In an optional implementation, the obtaining multi-band signal data in a substation communication network, performing fast Fourier transform to obtain spectrum distribution data, comprises:

[0014] Obtaining multi-band signal data in a substation communication network, performing frequency band division to obtain a frequency-division signal set;

[0015] According to the frequency-division signal set, performing windowing processing using a Hanning window function, and intercepting a signal segment with a sampling length of 1024 points to obtain a windowed signal segment;

[0016] According to the windowed signal segment, performing fast Fourier transform to obtain spectrum distribution data.

[0017] In an optional implementation, the according to the spectrum distribution data, calculating power spectral density ratio, performing frequency band division to obtain initial power ratio of main frequency band and interference frequency band, comprises:

[0018] According to the spectrum distribution data, performing beam analysis to obtain main lobe-side lobe characteristics;

[0019] According to the main lobe-side lobe characteristics, performing power spectral density calculation to obtain power spectral density ratio of main lobe and side lobe;

[0020] According to the power spectral density ratio and the spectrum distribution data, performing fundamental-harmonic separation, marking fundamental components as main frequency band and marking harmonic components as interference frequency band;

[0021] According to the main frequency band and the interference frequency band, frequency band power analysis is performed to obtain an initial power ratio of the main frequency band and the interference frequency band.

[0022] In an optional implementation, the adaptive filtering according to the initial power ratio and the multi-frequency band signal data includes:

[0023] According to the initial power ratio, spectrum leakage analysis is performed to determine a filter convergence step size;

[0024] According to the multi-frequency band signal data, fast Fourier transform is performed to extract initial fundamental wave amplitude fluctuation data;

[0025] According to the filter convergence step size and the initial fundamental wave amplitude fluctuation data, adaptive filtering is performed to obtain a fundamental wave amplitude fluctuation feature.

[0026] In an optional implementation, the wavelet decomposition according to the fundamental wave amplitude fluctuation feature includes:

[0027] According to the fundamental wave amplitude fluctuation feature, db4 wavelet base function is used for five-layer wavelet decomposition to generate a first high-frequency coefficient set;

[0028] According to the first high-frequency coefficient set, time-frequency energy distribution calculation is performed to generate a first time-frequency energy graph;

[0029] According to the first time-frequency energy graph and the first high-frequency coefficient set, least mean square optimization is performed to obtain a second high-frequency coefficient set;

[0030] According to the second high-frequency coefficient set, transient impact component analysis is performed to obtain an abnormal signal time-frequency energy distribution.

[0031] In an optional implementation, the K-means clustering division according to the abnormal signal time-frequency energy distribution includes:

[0032] According to the abnormal signal time-frequency energy distribution, a discrete frequency point set is extracted, amplitude and phase calculation is performed, and a first frequency point feature set is obtained;

[0033] According to the first frequency point feature set, Euclidean distance similarity calculation is performed to obtain a similarity matrix;

[0034] According to the similarity matrix and the first frequency point feature set, K-means clustering is used to divide the frequency points into a third harmonic group and a random noise group to obtain a first clustering result;

[0035] According to the first clustering result, a pre-established harmonic fault database is matched to obtain a fault type identifier.

[0036] In an optional embodiment, according to the fault type identifier and the abnormal signal time-frequency energy distribution, ring network topology propagation delay analysis is performed to locate a position coordinate of the fault occurrence and generate a fault diagnosis report, including:

[0037] According to the fault type identifier and the abnormal signal time-frequency energy distribution, a fault-related signal distribution is extracted, ring network topology propagation delay quantization processing is performed, and time delay distribution data is obtained.

[0038] According to the time delay distribution data, signal propagation path is decomposed segment by segment to obtain a node time delay variation.

[0039] According to the node time delay variation, corresponding voltage and current data are obtained and node impedance distribution data are calculated, and when the impedance distribution data are less than a preset impedance distribution threshold value, a node corresponding propagation path segment is marked as a preliminary fault area.

[0040] According to the preliminary fault area and the abnormal signal time-frequency energy distribution, fine comparison is performed to determine a position coordinate of the fault occurrence and generate a fault diagnosis report.

[0041] In a second aspect, the present application provides a fault diagnosis device for a substation communication network, including:

[0042] A data acquisition module is configured to acquire multi-frequency signal data in a substation communication network, perform fast Fourier transform, and obtain frequency spectrum distribution data.

[0043] A frequency band division module is configured to calculate a power spectrum density ratio according to the frequency spectrum distribution data, perform frequency band division, and obtain an initial power ratio of a main frequency band and an interference frequency band.

[0044] A dynamic filtering module is configured to perform adaptive filtering according to the initial power ratio and the multi-frequency signal data to obtain a fundamental wave amplitude fluctuation feature.

[0045] A wavelet decomposition module is configured to perform wavelet decomposition according to the fundamental wave amplitude fluctuation feature to obtain an abnormal signal time-frequency energy distribution.

[0046] A fault matching module is configured to perform K-means clustering division according to the abnormal signal time-frequency energy distribution, match a pre-established harmonic fault database, and obtain a fault type identifier.

[0047] The result output module is configured to perform ring network topology propagation delay analysis according to the fault type identification and the abnormal signal time-frequency energy distribution, locate the position coordinates where the fault occurs, and generate a fault diagnosis report.

[0048] In a third aspect, the present application also provides an electronic device comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the fault diagnosis method of the substation communication network according to any one of the above embodiments when executing the computer program.

[0049] In a fourth aspect, the present application also provides a computer readable storage medium comprising a stored computer program, wherein the computer readable storage medium controls the device where the computer readable storage medium is located to execute the fault diagnosis method of the substation communication network according to any one of the above embodiments when the computer program is running.

[0050] Compared with the prior art, the present application has the following beneficial effects:

[0051] (1) The present application obtains multi-band signal data in the substation communication network, performs frequency band division to obtain a frequency band division signal set, then performs windowing processing using the Hanning window function and intercepts a signal segment with a 1024-point sampling length, and finally performs fast Fourier transform to obtain frequency spectrum distribution data. This process effectively reduces unnecessary signal processing burden through segmented acquisition and frequency band division, concentrates on processing key data segments, can reduce computational complexity, thereby improving computational efficiency. The windowing processing of the Hanning window function reduces the frequency spectrum leakage phenomenon by smoothing the signal edge, ensures the clarity and stability of the frequency spectrum distribution data, and lays a solid foundation for subsequent frequency band analysis. Further, the setting of the 1024-point sampling length optimizes the resolution of data processing, reduces the amount of redundant calculation, and at the same time enhances the extraction accuracy of the frequency spectrum features, thereby improving the accuracy of the diagnosis step.

[0052] (2) The present application obtains the main lobe-sidelobe feature through beam analysis according to the frequency spectrum distribution data, then calculates the power spectrum density ratio of the main lobe and the sidelobe, further separates the fundamental wave and the harmonic wave and marks the main frequency band and the interference frequency band, and finally obtains the initial power ratio through frequency band power analysis. This process avoids the computational pressure of directly processing all frequency spectrum data by gradually filtering key feature information, can effectively shorten the analysis time and improve the computational efficiency. The beam analysis accurately distinguishes the main components of the signal by identifying the distribution characteristics of the main lobe and the sidelobe, and the calculation of the power spectrum density ratio further enhances the separation ability of the main frequency band and the interference frequency band. The combination of the fundamental wave-harmonic wave separation and the frequency band power analysis ensures the reliability of the initial power ratio, provides accurate data support for subsequent filtering and diagnosis, and thereby improves the accuracy of the entire process.

[0053] (3) The application determines the filter convergence step length by analyzing the spectrum leakage according to the initial power ratio, then extracts the initial fundamental amplitude fluctuation data by fast Fourier transform using multi-band signal data, and finally obtains the fundamental amplitude fluctuation feature by adaptive filtering combined with the filter convergence step length. This process optimizes the calculation process according to the spectrum leakage by dynamically adjusting the convergence step length, reduces unnecessary iterative operations, shortens the filtering time, and thus improves the calculation efficiency. Spectrum leakage analysis can identify the interference characteristics in the signal and guide the accurate adjustment of the filtering parameters, and the initial data extracted by fast Fourier transform provides a reliable basis for adaptive filtering. The final adaptive filtering process enhances the clarity of the fundamental amplitude fluctuation feature through targeted adjustment, ensuring the accuracy of subsequent wavelet decomposition.

[0054] (4) The application generates a first high-frequency coefficient set by five-layer wavelet decomposition using db4 wavelet base function according to the fundamental amplitude fluctuation feature, then calculates the first time-frequency energy distribution to generate a first time-frequency energy graph, obtains a second high-frequency coefficient set combined with the least mean square optimization, and finally analyzes the transient impact component to obtain the abnormal signal time-frequency energy distribution. This process focuses on high-frequency characteristics to reduce the amount of irrelevant data calculation and optimize the efficiency of the decomposition process. Five-layer decomposition of the db4 wavelet base function can effectively separate the high-frequency components in the signal, and the generated first high-frequency coefficient set provides accurate input for time-frequency energy distribution calculation. The least mean square optimization eliminates noise interference by adjusting the coefficients, further improving the purity of the second high-frequency coefficient set. Transient impact component analysis finally extracts the key features of abnormal signals, enhancing the clarity of the time-frequency energy distribution.

[0055] (5) The application extracts a discrete frequency point set according to the abnormal signal time-frequency energy distribution to obtain a first frequency point feature set by amplitude and phase calculation, then generates a similarity matrix by Euclidean distance similarity calculation, and uses K-means clustering to divide the frequency points and match the harmonic fault database to obtain the fault type identification. This process focuses on key frequency point information to reduce meaningless calculation and comparison, optimizing the efficiency of the clustering process. Euclidean distance similarity calculation generates a reliable similarity matrix by quantifying the differences between frequency points, providing accurate classification basis for K-means clustering, and the database matching of the clustering result further confirms the fault type. Such a process effectively distinguishes the third harmonic group and the random noise group, enhancing the accuracy of the fault type identification.

[0056] (6) The present invention extracts the fault-related signal distribution based on the fault type identification and the abnormal signal time-frequency energy distribution, performs ring network topology propagation delay quantification processing to obtain delay distribution data, then decomposes the signal propagation path section by section to obtain the delay change of each node, combines the impedance distribution data to mark the preliminary fault area, and finally performs refined comparison to determine the location coordinates and generate a fault diagnosis report. This process reduces redundant calculations by centrally processing key node data and optimizes the efficiency of path analysis. Delay quantification processing and path decomposition can accurately identify the time delay changes of each node, while the comparison of impedance distribution data further narrows the scope of the fault area. Refined comparison combined with abnormal signal characteristics and delay data ensures the accurate positioning of the location coordinates, and the fault diagnosis report finally generated is reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 1 is a flowchart of a fault diagnosis method for a substation communication network provided by a first embodiment of the present invention;

[0058] Figure 2 It is a structural diagram of a fault diagnosis system / device for a substation communication network provided by a second embodiment of the present invention. DETAILED DESCRIPTION

[0059] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0060] Reference Figure 1 A first embodiment of the present invention provides a method for diagnosing a fault in a substation communication network, comprising the following steps:

[0061] S11, acquiring multi-band signal data in the substation communication network, performing fast Fourier transform, and obtaining spectrum distribution data;

[0062] S12, calculating a power spectrum density ratio based on the spectrum distribution data, performing frequency band division, and obtaining an initial power ratio between a main frequency band and an interference frequency band;

[0063] S13, performing adaptive filtering based on the initial power ratio and the multi-band signal data to obtain fundamental wave amplitude fluctuation characteristics;

[0064] S14, performing wavelet decomposition according to the fundamental wave amplitude fluctuation characteristics to obtain the time-frequency energy distribution of the abnormal signal;

[0065] S15, performing K-means clustering based on the time-frequency energy distribution of the abnormal signal, matching the pre-established harmonic fault database, and obtaining a fault type identifier;

[0066] S16, performing a ring network topology propagation delay analysis based on the fault type identifier and the abnormal signal time-frequency energy distribution, locating the position coordinates of the fault, and generating a fault diagnosis report.

[0067] In step S11, it is necessary to obtain multi-band signal data in the substation communication network, perform fast Fourier transform, and obtain spectrum distribution data.

[0068] In one implementation, multi-band signal data in a substation communication network is acquired and fast Fourier transformed to obtain spectrum distribution data, including:

[0069] Acquire multi-band signal data in the substation communication network, perform frequency band division, and obtain a sub-band signal set; perform windowing processing using a Hanning window function based on the sub-band signal set, intercept signal segments with a sampling length of 1024 points, and obtain windowed signal segments; perform fast Fourier transform on the windowed signal segments to obtain spectrum distribution data.

[0070] It should be noted that the specific implementation of obtaining multi-band signal data in the substation communication network, performing frequency band division, and obtaining a frequency band signal set is to collect original signal data containing multiple frequency components from the network through a dedicated signal acquisition device, and then based on the preset frequency range division rule, such as dividing the signal into categories such as the power frequency band and the harmonic band, it is divided into multiple independent frequency band signal sets to facilitate subsequent targeted processing. According to the frequency band signal set, the Hanning window function is used for windowing processing, and a signal segment with a sampling length of 1024 points is intercepted. The implementation of obtaining the windowed signal segment is to apply a smoothing window to each frequency band signal by applying the Hanning window function to reduce the sudden change effect of the signal edge, and then intercept a fixed length of 1024 sampling point data segment to ensure the integrity and consistency of the signal segment. The Hanning window function is a smoothing windowing method that can effectively reduce spectrum leakage. According to the windowed signal segment, a fast Fourier transform is performed to obtain the spectrum distribution data by applying a fast Fourier transform algorithm to the windowed signal segment, converting the time domain signal into a frequency domain representation, extracting the frequency component, amplitude and phase information, and generating spectrum distribution data. The spectrum distribution data reflects the energy distribution of the signal at different frequencies. It is obtained through the above-mentioned windowing and transformation process and can be used for subsequent power spectrum density ratio calculation and frequency band analysis to identify the main frequency band and the interference frequency band.

[0071] In step S12, it is necessary to calculate the power spectrum density ratio according to the spectrum distribution data, perform frequency band division, and obtain the initial power ratio of the main frequency band to the interference frequency band.

[0072] In one implementation, calculating the power spectrum density ratio based on the spectrum distribution data, performing frequency band division, and obtaining an initial power ratio between the main frequency band and the interference frequency band includes:

[0073] Based on the spectrum distribution data, beam analysis is performed to obtain main lobe-side lobe characteristics; based on the main lobe-side lobe characteristics, power spectrum density calculation is performed to obtain the power spectrum density ratio of the main lobe and the side lobe; based on the power spectrum density ratio and the spectrum distribution data, fundamental-harmonic separation is performed, and the fundamental component is marked as the main frequency band, and the harmonic component is marked as the interference frequency band; based on the main frequency band and the interference frequency band, frequency band power analysis is performed to obtain the initial power ratio of the main frequency band to the interference frequency band.

[0074] It should be noted that, according to the spectrum distribution data, beam analysis is performed to obtain the specific implementation operations of the main lobe-side lobe characteristics as follows: First, according to the generated spectrum distribution data, beam analysis is performed, and the spectrum data is divided into multiple sub-bands according to the frequency axis. The specific basis is adaptive bandwidth division based on signal energy distribution, that is, the sub-band is divided according to the inflection point of the spectrum energy change to ensure that each sub-band contains continuous frequency components. The sub-band width is dynamically adjusted according to the energy gradient change, and the signal intensity peak in each sub-band is calculated. The peak with the highest amplitude is identified as the main lobe, and the side lobe is defined as the secondary peak on both sides of the main lobe with an amplitude lower than the main lobe but still higher than the noise floor, such as a peak with an amplitude 6dB lower than the main lobe peak but 3dB higher than the noise floor; then, the center frequency and amplitude of the main lobe, as well as the distribution range and relative amplitude of the side lobe are recorded, and the main lobe-side lobe characteristic data file is output for use in subsequent steps. The power spectrum density is calculated according to the main lobe-side lobe characteristics, and the specific implementation operation of obtaining the power spectrum density ratio of the main lobe to the side lobe is as follows: according to the main lobe-side lobe characteristic data, the energy density of the main lobe and the side lobe in their respective frequency ranges is calculated, the amplitude data of the main lobe and the side lobe are squared to obtain the power value, and then the power value is divided by the corresponding frequency bandwidth to generate the power spectrum density of each part; thereafter, the power spectrum density of the main lobe and the side lobe is compared, the ratio between the two is calculated, and the result is recorded as the power spectrum density ratio.

[0075] According to the power spectrum density ratio and the spectrum distribution data, the fundamental-harmonic separation is performed, the fundamental component is marked as the main frequency band, and the harmonic component is marked as the interference frequency band. The specific implementation operation is as follows: According to the spectrum distribution data and the power spectrum density ratio, the preset frequency band judgment threshold is obtained according to the statistical analysis of the historical substation communication network signal, specifically by analyzing the distribution law of the main lobe and side lobe power spectrum density under a large number of normal operating conditions, and selecting the typical ratio range of the main frequency band energy greater than the interference frequency band, such as a value of 10 decibels, as the benchmark threshold for distinguishing the main frequency band from the interference frequency band. When the power spectrum density ratio is greater than the preset frequency band judgment threshold, the frequency band corresponding to the main lobe is judged. The frequency component is the main frequency band, and the frequency component corresponding to the side lobe is the interference frequency band. When the power spectrum density ratio is less than the preset frequency band judgment threshold, the frequency components corresponding to the main lobe and side lobe are marked as interference frequency bands. The frequency band with a power spectrum density ratio lower than the threshold is identified, smoothing filtering is performed, and the power spectrum density ratio is recalculated. After identifying the frequency band with a power spectrum density ratio lower than the threshold, a Gaussian smoothing filter is used to suppress noise in the band, and the main lobe recognition range is adjusted to expand it to the peak area after the energy distribution is smoothed. At the same time, the side lobe judgment rule is modified, and the peak with an amplitude lower than 6dB of the main lobe is re-included in the analysis. The power spectrum density ratio of the main lobe and the side lobe is iteratively calculated until the ratio is stable. Divide the main lobe and side lobe; then, traverse the spectrum distribution data, identify the fundamental component and its multiple harmonic components according to the frequency multiple relationship, and specifically extract the lowest frequency peak from the spectrum data as the fundamental component, and sequentially detect whether there is a peak with significant amplitude at its double, triple and other integer multiple frequencies. The peak amplitude must be 5 decibels higher than the noise floor and reach a level of more than 50% of the fundamental amplitude to be determined as a significant peak to ensure the accuracy of harmonic component identification. If it exists, it is marked as a harmonic component, the fundamental component is marked as the main frequency band, and the harmonic component is marked as the interference frequency band. Generate a segmented marking file to record the frequency range of the main frequency band and the interference frequency band.

[0076] The specific implementation steps for performing frequency band power analysis based on the primary and interference bands to obtain the initial power ratio of the primary and interference bands are as follows: Based on the segmented marker file, the frequency ranges of the primary and interference bands are extracted, the amplitude data within each band is cumulatively summed, and the total power of each band is calculated. Next, the total power of the primary band is compared with the total power of the interference band, and the energy ratio between the two is calculated to generate the initial power ratio of the primary band to the interference band. This ratio is output as a percentage or decibel and stored as input data for subsequent adaptive filtering. The initial power ratio of the primary band to the interference band is a quantitative indicator that describes the energy of the primary band relative to the energy of the interference band and can be used to guide filter parameter adjustment and abnormal signal identification.

[0077] In step S13, it is necessary to perform adaptive filtering based on the initial power ratio and the multi-band signal data to obtain fundamental wave amplitude fluctuation characteristics.

[0078] In an implementation manner, the fundamental amplitude fluctuation feature is obtained by adaptive filtering according to the initial power ratio and the multi-band signal data.

[0079] The filter convergence step is determined by spectrum leakage analysis according to the initial power ratio, the initial fundamental amplitude fluctuation data is extracted by performing fast Fourier transform on the multi-band signal data, and the fundamental amplitude fluctuation feature is obtained by adaptive filtering according to the filter convergence step and the initial fundamental amplitude fluctuation data.

[0080] It should be noted that the filter convergence step is determined by spectrum leakage analysis according to the initial power ratio, which analyzes the energy distribution of the main frequency band and the interference frequency band by power spectrum density comparison, detects whether there is a leakage phenomenon in the spectrum, first sets a leakage detection window according to the initial power ratio, for example, takes the fundamental frequency as the center and covers ±5% frequency bandwidth, adjusts the window step to be an integer multiple of the frequency resolution, such as moving once every 0.1 Hz, and calculates the fluctuation range of the power ratio in the window; then, the power ratio data is smoothed by moving average, and the abnormal points exceeding the preset amplitude of the average value are identified as leakage indicators, the preset amplitude is obtained according to the statistical analysis of historical transformer station signal samples, specifically, the standard deviation of the power ratio fluctuation under normal operating conditions is calculated, for example, the average value plus twice the standard deviation is taken as the preset amplitude threshold, for example, if the standard deviation is 0.5 decibels, the preset amplitude is the average value plus 1 decibel; finally, the filter convergence step is dynamically generated according to the leakage indicator, and the step value is determined by being inversely proportional to the reciprocal of the power ratio, for example, the step is gradually reduced when the power ratio decreases, to ensure the stability of the filtering process, and the filter convergence step is generated to optimize the subsequent filtering.

[0081] The initial fundamental amplitude fluctuation data is extracted by performing fast Fourier transform on the multi-band signal data, which realizes the conversion from time domain to frequency domain by fast Fourier transform, divides the continuous time period according to the multi-band signal data, and ensures that each segment contains a complete period, specifically, first, the interval time of the signal zero-crossing point is identified by zero-crossing detection according to the multi-band signal data, and the period length of the fundamental signal is determined, then the continuous time period is divided based on the period length, and the number of sampling points in each segment is ensured to be an integer multiple of the period length, for example, if the period is 20 ms and the sampling rate is 50 kHz, each segment contains 1000 sampling points, and finally the starting point between segments is adjusted to ensure that each time period contains a complete period, avoiding signal truncation; then, the time domain signal is converted to the frequency domain by applying fast Fourier transform, the main amplitude peak in the fundamental frequency range is identified, and then the sequence of the fundamental amplitude change over time is extracted, which is recorded as the initial fundamental amplitude fluctuation data for subsequent filtering processing.

[0082] Adaptive filtering is performed based on the filter convergence step size and the initial fundamental amplitude fluctuation data to obtain the fundamental amplitude fluctuation characteristics. This step dynamically adjusts the filter parameters through minimum mean square error adaptive filtering. The filter coefficients are initialized based on the filter convergence step size and the initial fundamental amplitude fluctuation data. The iteration step size is controlled based on the convergence step size, and the filter coefficients are gradually adjusted to minimize the error. After filtering out interference noise, the smoothed fundamental amplitude fluctuation characteristics are output. The fundamental amplitude fluctuation characteristics are continuous data reflecting the change in the amplitude of the fundamental signal over time, which can be used to detect the time-frequency characteristics of abnormal signals in subsequent wavelet decomposition.

[0083] In step S14, it is necessary to perform wavelet decomposition according to the fundamental wave amplitude fluctuation characteristics to obtain the time-frequency energy distribution of the abnormal signal.

[0084] In one implementation, wavelet decomposition is performed based on the fundamental wave amplitude fluctuation characteristics to obtain the time-frequency energy distribution of the abnormal signal, including:

[0085] According to the fundamental amplitude fluctuation characteristics, the db4 wavelet basis function is used to perform five-layer wavelet decomposition to generate a first high-frequency coefficient set; based on the first high-frequency coefficient set, the time-frequency energy distribution is calculated to generate a first time-frequency energy graph; based on the first time-frequency energy graph and the first high-frequency coefficient set, the minimum mean square optimization is performed to obtain a second high-frequency coefficient set; based on the second high-frequency coefficient set, the transient impact component analysis is performed to obtain the time-frequency energy distribution of the abnormal signal.

[0086] It should be noted that, according to the fundamental wave amplitude fluctuation characteristics, the db4 wavelet basis function is used to perform five-layer wavelet decomposition to generate the first high-frequency coefficient set. This step uses the db4 wavelet basis function to perform five-layer decomposition of the fundamental wave amplitude fluctuation characteristic data. According to the fundamental wave amplitude fluctuation characteristic data, the number of decomposition layers is set to five, and the db4 wavelet basis function is applied to perform high and low frequency separation to extract high-frequency detail coefficients; then, the low-frequency part is decomposed in the lower layer in turn, and the high-frequency coefficients of each decomposition are retained to generate the first high-frequency coefficient set file to record the detailed changes in the signal.

[0087] According to the first set of high-frequency coefficients, the time-frequency energy distribution is calculated to generate a first time-frequency energy map. This step involves dividing the first set of high-frequency coefficients into fixed time windows, such as one window every 0.1 seconds, through short-time Fourier transform. Specifically, the discrete high-frequency coefficient sequence is time-frequency mapped according to a 0.1-second time window, and the energy of the high-frequency coefficient in each window is calculated. The amplitudes of the coefficients in the window are squared and summed up. Then, the energy values ​​are normalized, and a two-dimensional time-frequency energy distribution matrix is ​​generated based on the total window energy divided by the maximum energy value. Finally, the matrix data is plotted as a first time-frequency energy map, with the horizontal axis being time, the vertical axis being frequency, and the color representing energy intensity.

[0088] Reflects the change of signal energy over time and frequency. Based on the first time-frequency energy map and the first high-frequency coefficient set, the minimum mean square optimization is performed to obtain the second high-frequency coefficient set. This step is optimized by the minimum mean square error. According to the first time-frequency energy map, the energy distribution is scanned, and areas where the energy values ​​deviate from the average level are identified as abnormal areas. Then, the first high-frequency coefficient set is extended to the same two-dimensional time-frequency grid as the time-frequency energy map by interpolation, and the sum of the squares of the point product differences between the two is calculated as the fitting error, and then the high-frequency coefficient weights are adjusted. The gradient descent method is used to gradually adjust the weight of each coefficient in the high-frequency coefficient set according to the gradient direction of the fitting error. The step size is dynamically set according to the error change rate. For example, the step size is halved when the error decreases. The iteration stops after 100 times or when the error is less than 0.01 to generate the second high-frequency coefficient set to enhance the accuracy of signal details.

[0089] Based on the second set of high-frequency coefficients, transient impact component analysis is performed to obtain the abnormal signal's time-frequency energy distribution. This step first sets an amplitude change threshold based on the second set of high-frequency coefficients. Based on previously acquired historical data statistics, the time-frequency energy distribution is calculated when the amplitude difference between adjacent sampling points in the second set of high-frequency coefficients exceeds the amplitude change threshold, generating the abnormal signal's time-frequency energy distribution. The abnormal signal's time-frequency energy distribution represents the energy distribution of the abnormal signal in both time and frequency dimensions and can be used for subsequent K-means clustering to classify fault types.

[0090] In step S15, it is necessary to perform K-means clustering based on the time-frequency energy distribution of the abnormal signal, match it with a pre-established harmonic fault database, and obtain a fault type identifier.

[0091] In one implementation, K-means clustering is performed based on the time-frequency energy distribution of the abnormal signal, and a pre-established harmonic fault database is matched to obtain a fault type identification, including:

[0092] According to the time-frequency energy distribution of the abnormal signal, a discrete frequency point set is extracted, and amplitude and phase calculations are performed to obtain a first frequency point feature set; based on the first frequency point feature set, Euclidean distance similarity calculation is performed to obtain a similarity matrix; based on the similarity matrix and the first frequency point feature set, K-means clustering is used to divide the frequency points into a third harmonic group and a random noise group to obtain a first clustering result; based on the first clustering result, a pre-established harmonic fault database is matched to obtain a fault type identification.

[0093] It should be noted that, according to the time-frequency energy distribution of the abnormal signal, a discrete frequency point set is extracted, and amplitude and phase calculations are performed to obtain the first frequency point feature set. This step first identifies the local maximum point in the energy distribution as a discrete frequency point based on the time-frequency energy distribution of the abnormal signal, and then calculates the amplitude of each frequency point, calculates the signal energy through the square root, and obtains the phase by extracting the frequency domain phase spectrum through the fast Fourier transform. Specifically, according to the time-frequency energy distribution of the abnormal signal, the fast Fourier transform is applied to convert the time domain signal segment corresponding to each discrete frequency point, extract the phase spectrum in the frequency domain, and generate the first frequency point feature set file.

[0094] Based on the first frequency point feature set, Euclidean distance similarity calculation is performed to obtain a similarity matrix. This step first extracts all frequency point data from the first frequency point feature set. Each frequency point contains two attributes, amplitude and phase, which are recorded as a two-dimensional vector. The amplitude is directly recorded as a positive value. The phase value may be directly used for Euclidean distance calculation due to its periodicity (0~2π cycle), which may lead to boundary errors. For example, the difference in phase from 359° to 1° is mistakenly judged as a large distance. To solve this problem, the phase value is first converted into sine and cosine components (such as cos(phase), sin(phase)) before being recorded as a two-dimensional vector to form a three-dimensional vector (amplitude, cos(phase), sin(phase)) to ensure the influence of phase periodicity on distance calculation. Minimize, thereby improving the accuracy of similarity calculation; then, traverse all possible frequency point pairs, and for each pair of frequency points, calculate the amplitude difference and phase difference; then, calculate the square sum of the amplitude difference and phase difference between each pair of frequency points; then, take the square root of the square sum to calculate the Euclidean distance between the two points; finally, to avoid the distance being 0 resulting in infinity, add a constant (such as 0.1) to each Euclidean distance value, and then take its reciprocal to obtain the similarity value. The larger the similarity value, the more similar the two frequency points are; fill these similarity values ​​into an N×N matrix, where N is the number of frequency points, the matrix diagonal elements are 1 (the similarity between itself and itself is the highest), and the non-diagonal elements represent the similarity between the two frequency points, thereby generating a similarity matrix.

[0095] According to the similarity matrix and the first frequency feature set, K-means clustering is used to divide the frequency points into the third harmonic group and the random noise group to obtain the first clustering result. In this step, K-means clustering is performed. First, the number of clusters is set to 2, representing the third harmonic group and the random noise group respectively; then, the cluster center is initialized based on the similarity matrix, and the distance from each frequency point to the cluster center is iteratively calculated and assigned to the nearest cluster until convergence. If the frequency multiple concentration in the clustering result meets the preset standard, for example, the frequency points with frequency multiples of three times or less account for more than 50%, then it is classified as the third harmonic group, otherwise it is classified as the random noise group, and the first clustering result is generated, and the classification of each frequency point is recorded.

[0096] Based on the first clustering results, a pre-established harmonic fault database is matched. This database, constructed by collecting actual operating data from 500 substations, covers 10 common harmonic fault types, including transformer core saturation and line resonance, with over 2,000 data samples. Feature dimensions include frequency amplitude, phase information, and spectral shape. These features are extracted through Fourier transform and energy distribution analysis to ensure that the database contains comprehensive fault feature patterns for matching and obtaining a fault type identifier. This step involves feature matching, extracting the frequency features of the third harmonic group based on the first clustering results. These features are then compared with standard feature patterns in the pre-established harmonic fault database. The best match is selected based on similarity to generate a fault type identifier. The fault type identifier is a code that indicates a specific fault category (such as transformer core saturation) and can be used for subsequent ring network topology analysis to locate the fault.

[0097] In step S16, it is necessary to perform a ring network topology propagation delay analysis based on the fault type identifier and the abnormal signal time-frequency energy distribution, locate the position coordinates of the fault, and generate a fault diagnosis report.

[0098] In one implementation, based on the fault type identifier and the time-frequency energy distribution of the abnormal signal, a ring network topology propagation delay analysis is performed to locate the coordinates of the fault location and generate a fault diagnosis report, including:

[0099] Based on the fault type identification and the time-frequency energy distribution of the abnormal signal, the fault-related signal distribution is extracted, and the ring network topology propagation delay is quantified to obtain delay distribution data; based on the delay distribution data, the signal propagation path is decomposed section by section to obtain the delay change of each node; based on the delay change of each node, the corresponding voltage and current data are obtained and the impedance distribution data of each node is calculated. When the impedance distribution data is less than a preset impedance distribution threshold, the propagation path section corresponding to the node is marked as a preliminary fault area; based on the preliminary fault area and the time-frequency energy distribution of the abnormal signal, a detailed comparison is performed to determine the location coordinates of the fault, and a fault diagnosis report is generated.

[0100] It should be noted that, according to the fault type identification and the abnormal signal time-frequency energy distribution, the fault-related signal distribution is extracted, and the ring network topology propagation delay is quantified to obtain the delay distribution data. This step first extracts the high-energy frequency band related to the fault type from the abnormal signal time-frequency energy distribution according to the fault type identification to generate the fault-related signal distribution. Then, combined with the ring network topology diagram, the signal transmission path between each node is recorded, and the propagation time difference of the signal from the source node to each node is calculated to generate the delay distribution data. According to the delay distribution data, the signal propagation path is decomposed section by section to obtain the delay change of each node. This step is achieved by first dividing the propagation segments between each node pair of the ring network according to the delay distribution data, calculating the delay change of each signal transmission segment, recording the delay increment between each node, and generating the delay change of each node.

[0101] Based on the delay changes at each node, corresponding voltage and current data are obtained and impedance distribution data for each node is calculated. When the impedance distribution data is less than a preset impedance distribution threshold, the corresponding propagation path segment of the node is marked as a preliminary fault area. This step first obtains voltage and current data for each node based on the delay changes at each node and the network topology. The impedance value is calculated by dividing the voltage by the current. The preset impedance distribution threshold is determined by subtracting two standard deviations from the average impedance value of historical normal operation data. The impedance value of each node in the impedance distribution data is compared with the preset impedance distribution threshold. When the node impedance value is less than the preset impedance distribution threshold, the corresponding propagation path segment of the node is marked as a preliminary fault area. Based on the preliminary fault area and the abnormal signal time-frequency energy distribution, a detailed comparison is performed to determine the location coordinates of the fault, and a fault diagnosis report is generated. This step extracts the energy peak of the corresponding area in the abnormal signal time-frequency energy distribution based on the preliminary fault area. Combining the topological path and energy distribution, the energy fluctuation amplitude of each node is calculated. The node with the largest energy fluctuation amplitude is selected as the fault location coordinate, and a fault diagnosis report is generated. The fault diagnosis report is a document that records the fault type and location coordinates and can be used as a guide for network maintenance and fault repair.

[0102] In summary, the present invention discloses a fault diagnosis method for a substation communication network. The present invention obtains multi-band signal data and performs spectrum analysis, uses adaptive filtering and wavelet decomposition to extract abnormal signal features, and combines K-means clustering and ring network topology delay analysis to locate faults, thereby achieving accurate diagnosis of network faults in the face of multi-band complex signal interference.

[0103] Reference Figure 2 A second embodiment of the present invention provides a fault diagnosis device for a substation communication network, comprising:

[0104] The data acquisition module is configured to acquire multi-band signal data in a substation communication network, and perform fast Fourier transform to obtain frequency spectrum distribution data.

[0105] The frequency band division module is configured to calculate a power spectrum density ratio according to the frequency spectrum distribution data, perform frequency band division, and obtain an initial power ratio of a main frequency band and an interference frequency band.

[0106] The dynamic filtering module is configured to perform adaptive filtering according to the initial power ratio and the multi-band signal data, and obtain a fundamental amplitude fluctuation feature.

[0107] The wavelet decomposition module is configured to perform wavelet decomposition according to the fundamental amplitude fluctuation feature, and obtain abnormal signal time-frequency energy distribution.

[0108] The fault matching module is configured to perform K-means clustering division according to the abnormal signal time-frequency energy distribution, match a pre-established harmonic fault database, and obtain a fault type identifier.

[0109] The result output module is configured to perform ring network topology propagation delay analysis according to the fault type identifier and the abnormal signal time-frequency energy distribution, locate a position coordinate of fault occurrence, and generate a fault diagnosis report.

[0110] It should be noted that the substation communication network fault diagnosis device provided by the embodiments of the present application is used to execute all process steps of the substation communication network fault diagnosis method provided by the above embodiments, and the working principles and advantages of the two are one-to-one corresponding, thus no longer being described in detail.

[0111] The embodiments of the present application further provide an electronic device. The electronic device comprises a processor, a memory, and a computer program stored in the memory and capable of running on the processor, for example, a substation communication network fault diagnosis program. The processor executes the computer program to implement the steps in the above various substation communication network fault diagnosis method embodiments, for example Figure 1 the step S11 shown. Alternatively, the processor executes the computer program to implement the functions of the modules / units in the above various device embodiments, for example, a fault matching module.

[0112] For example, the computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present application. The one or more modules / units can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the electronic device.

[0113] The electronic device can be a computing device such as a desktop computer, a notebook computer, a palm computer, a smart tablet, etc. The electronic device can include, but is not limited to, a processor, a memory. Those skilled in the art can understand that the above components are only examples of the electronic device and do not constitute a limitation on the electronic device, and the electronic device can include more or fewer components than the above, or combine certain components, or different components, for example, the electronic device can also include an input / output device, a network access device, a bus, etc.

[0114] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The processor is the control center of the electronic device, and connects various parts of the electronic device through various interfaces and lines.

[0115] The memory can be used to store the computer program and / or modules, and the processor realizes various functions of the electronic device by running or executing the computer program and / or modules stored in the memory, and calling data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required by a function (such as a sound playing function, an image playing function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, for example, a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state memory device.

[0116] The modules / units integrated in the electronic device, if realized in the form of software function units and sold or used as independent products, can be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. The computer program can implement the steps of each method embodiment when executed by a processor. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or some intermediate forms, etc. The computer readable medium can include any entity or device, recording medium, U disk, 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. that can carry the computer program code. It should be noted that the contents included in the computer readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunication signals.

[0117] It should be noted that the above-described device embodiments are only schematic, and the units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or distributed on multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the embodiment. In addition, the connection relationship between the modules in the device embodiment provided by the present application indicates that there is a communication connection between them, which can be realized as one or more communication buses or signal lines. Those skilled in the art can understand and implement it without creative labor.

[0118] The above-described specific embodiments further illustrate the purpose, technical solutions and beneficial effects of the present application. It should be understood that the above-described specific embodiments are only for the specific embodiments of the present application and are not used to limit the protection scope of the present application. It is particularly pointed out that any modification, equivalent replacement, improvement, etc. made by those skilled in the art within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A fault diagnosis method for a substation communication network, characterized in that: include: Acquire multi-band signal data from the substation communication network, perform fast Fourier transform, and obtain spectrum distribution data; Calculating the power spectrum density ratio based on the spectrum distribution data, performing frequency band division, and obtaining an initial power ratio between the main frequency band and the interference frequency band; Performing adaptive filtering based on the initial power ratio and the multi-band signal data to obtain fundamental wave amplitude fluctuation characteristics; According to the fundamental amplitude fluctuation characteristics, wavelet decomposition is performed to obtain the time-frequency energy distribution of the abnormal signal; According to the time-frequency energy distribution of the abnormal signal, K-means clustering is performed, and a pre-established harmonic fault database is matched to obtain a fault type identification; Performing ring network topology propagation delay analysis based on the fault type identifier and the abnormal signal time-frequency energy distribution, locating the location coordinates of the fault, and generating a fault diagnosis report; The step of calculating the power spectrum density ratio based on the spectrum distribution data, dividing the frequency bands, and obtaining the initial power ratio between the main frequency band and the interference frequency band includes: Performing beam analysis based on the spectrum distribution data to obtain mainlobe-sidelobe characteristics; Calculating the power spectrum density based on the main lobe-side lobe characteristics to obtain a power spectrum density ratio of the main lobe to the side lobe; Performing fundamental-harmonic separation according to the power spectrum density ratio and the spectrum distribution data, marking the fundamental component as a main frequency band, and marking the harmonic component as an interference frequency band; Performing frequency band power analysis based on the main frequency band and the interference frequency band to obtain an initial power ratio of the main frequency band to the interference frequency band; The step of performing adaptive filtering based on the initial power ratio and the multi-band signal data to obtain fundamental wave amplitude fluctuation characteristics includes: Performing spectrum leakage analysis based on the initial power ratio to determine a filter convergence step size; Performing a fast Fourier transform on the multi-band signal data to extract initial fundamental wave amplitude fluctuation data; Adaptive filtering is performed according to the filtering convergence step size and the initial fundamental wave amplitude fluctuation data to obtain fundamental wave amplitude fluctuation characteristics.

2. The fault diagnosis method for a substation communication network according to claim 1, characterized in that: The method of acquiring multi-band signal data in the substation communication network and performing fast Fourier transform to obtain spectrum distribution data includes: Obtain multi-band signal data in the substation communication network, divide the frequency bands, and obtain a sub-band signal set; Performing windowing processing on the frequency-band signal set using a Hanning window function to intercept a signal segment with a sampling length of 1024 points to obtain a windowed signal segment; A fast Fourier transform is performed on the windowed signal segment to obtain spectrum distribution data.

3. The fault diagnosis method for a substation communication network according to claim 1, characterized in that: The step of performing wavelet decomposition according to the fundamental wave amplitude fluctuation characteristics to obtain the time-frequency energy distribution of the abnormal signal includes: According to the fundamental amplitude fluctuation characteristics, a five-layer wavelet decomposition is performed using the db4 wavelet basis function to generate a first high-frequency coefficient set; performing time-frequency energy distribution calculation based on the first high-frequency coefficient set to generate a first time-frequency energy map; performing least mean square optimization based on the first time-frequency energy map and the first high-frequency coefficient set to obtain a second high-frequency coefficient set; According to the second high-frequency coefficient set, transient impact component analysis is performed to obtain the time-frequency energy distribution of the abnormal signal.

4. The fault diagnosis method for a substation communication network according to claim 1, characterized in that: The method of performing K-means clustering based on the time-frequency energy distribution of the abnormal signal and matching the pre-established harmonic fault database to obtain a fault type identifier includes: Extracting a discrete frequency point set according to the time-frequency energy distribution of the abnormal signal, performing amplitude and phase calculations to obtain a first frequency point feature set; Performing Euclidean distance similarity calculation based on the first frequency feature set to obtain a similarity matrix; Based on the similarity matrix and the first frequency feature set, K-means clustering is used to divide the frequency points into a third harmonic group and a random noise group to obtain a first clustering result; According to the first clustering result, a pre-established harmonic fault database is matched to obtain a fault type identifier.

5. The fault diagnosis method for a substation communication network according to claim 1, characterized in that: The method of performing a ring network topology propagation delay analysis based on the fault type identifier and the abnormal signal time-frequency energy distribution, locating the location coordinates of the fault, and generating a fault diagnosis report includes: Extracting fault-related signal distribution according to the fault type identifier and the abnormal signal time-frequency energy distribution, performing ring network topology propagation delay quantification processing, and obtaining delay distribution data; Decomposing the signal propagation path section by section based on the delay distribution data to obtain the delay variation of each node; According to the delay changes of each node, the corresponding voltage and current data are obtained and the impedance distribution data of each node is calculated. When the impedance distribution data is less than a preset impedance distribution threshold, the propagation path section corresponding to the node is marked as a preliminary fault area; Based on the preliminary fault area and the time-frequency energy distribution of the abnormal signal, a detailed comparison is performed to determine the location coordinates of the fault and generate a fault diagnosis report.

6. A fault diagnosis system for a substation communication network, used to implement the method described in any one of claims 1 to 5, characterized in that: include: The data acquisition module is used to obtain multi-band signal data in the substation communication network, perform fast Fourier transform, and obtain spectrum distribution data; A frequency band division module is used to calculate the power spectrum density ratio according to the spectrum distribution data, perform frequency band division, and obtain an initial power ratio between the main frequency band and the interference frequency band; a dynamic filtering module, configured to perform adaptive filtering based on the initial power ratio and the multi-band signal data to obtain fundamental wave amplitude fluctuation characteristics; A wavelet decomposition module is used to perform wavelet decomposition according to the fundamental amplitude fluctuation characteristics to obtain the time-frequency energy distribution of the abnormal signal; A fault matching module is used to perform K-means clustering based on the time-frequency energy distribution of the abnormal signal, match it with a pre-established harmonic fault database, and obtain a fault type identification; The result output module is used to perform ring network topology propagation delay analysis based on the fault type identifier and the abnormal signal time-frequency energy distribution, locate the position coordinates of the fault, and generate a fault diagnosis report.

7. An electronic device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the method for diagnosing a fault in a substation communication network according to any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the fault diagnosis method for the substation communication network according to any one of claims 1 to 5.

Images