A method for judging short-circuit fault in distribution network based on the variation of phase voltage amplitude at load end

Abstract

The present invention discloses a method for judging a short-circuit fault in a distribution network based on a phase voltage amplitude variation at a load end, comprising the following steps: S1, installing a voltage monitoring device at the load end, collecting phase voltage data in real time, and generating a voltage amplitude variation; S2, preprocessing the data, eliminating noise interference, and generating preprocessed voltage variation data; S3, using a high-order statistical spectrum subtraction method to analyze the preprocessed data, identifying abnormal components in the signal, and preliminarily judging a short-circuit fault; S4, confirming and recording the abnormal position through a quantum tunneling effect detection algorithm; S5, using a generalized maximum entropy principle to perform time series analysis and confirm the fault; S6, accurately locating the fault point based on the zero scattering theory; S7, classifying the fault type according to the minimum entropy productivity principle; S8, dynamically adjusting the fault isolation scheme based on generalized predictive control. The present invention improves the accuracy and response speed of fault detection, and is suitable for fault diagnosis and processing of complex power systems.

Description

Technical Field

[0001] The present invention relates to the technical field of power system fault detection and diagnosis, and in particular to a distribution network short-circuit fault judgment method based on load-end phase voltage amplitude variation. Background Art

[0002] As the complexity of power systems continues to increase, especially in modern distribution networks, the detection and location of short-circuit faults has become a key issue for the stable operation of power systems. Traditional fault detection methods rely on single voltage or current waveform analysis, usually by monitoring current mutations, voltage drops and other phenomena to determine the occurrence of faults. However, these methods have significant limitations when dealing with complex power grid environments, mainly manifested in the following aspects:

[0003] 1. Low detection accuracy: Traditional fault detection methods are mostly based on waveform analysis. When dealing with complex power grid environments, these methods are easily affected by noise interference and various fault types, resulting in low detection accuracy. For example, a small fluctuation in the voltage waveform may be misjudged as a fault, or the real fault signal may be masked by noise, resulting in frequent missed or false alarms.

[0004] 2. Slow response: Traditional methods rely on manual judgment or static algorithms for fault analysis, which makes it impossible for the system to respond in time when a fault occurs. Especially in complex distribution networks, fault signals spread quickly, and traditional methods have a slow response speed and cannot accurately locate and handle faults in the first time of the fault, which may trigger a chain reaction and lead to a larger range of power outages.

[0005] 3. Difficulty in handling multi-point faults: As the scale of the power grid expands, multi-point faults are becoming more and more common. However, existing fault detection methods can only handle single fault points and cannot effectively deal with complex scenarios of multi-point faults. This results in the inability of traditional methods to accurately locate all fault points when multi-point faults occur, and may even result in misjudgment, affecting the recovery of the power grid and the handling of the fault.

[0006] 4. Lack of multi-dimensional data analysis capabilities: Existing technologies usually only use a single data dimension in fault detection, ignoring other factors that may have an important impact on fault judgment. This single-dimensional analysis method limits the comprehensiveness and accuracy of fault detection, especially when facing multiple disturbances and abnormal conditions in a complex power grid environment, it is difficult to provide accurate fault location and type discrimination.

[0007] 5. Unable to handle complex nonlinear phenomena: Fault signals in power grids are usually highly nonlinear and time-varying, and traditional linear analysis methods are often unable to cope with these complex nonlinear phenomena. For example, the nonlinear characteristics in fault signals may cause traditional methods to be unable to accurately capture key fault information, thus affecting the overall detection effect.

[0008] 6. Lack of dynamic adaptability: The power grid operating environment is complex and changeable. Traditional methods usually rely on preset thresholds or rules for fault detection. These thresholds or rules often cannot adapt to the dynamically changing power grid environment in actual applications. For example, when the load changes frequently or the power grid structure is adjusted, the fixed thresholds in the traditional method may lead to inaccurate detection results and cannot effectively cope with the dynamic changes of the power grid.

[0009] 7. Reliance on human intervention: In many cases, traditional fault detection methods rely on human intervention and empirical judgment, which not only increases the time cost of fault handling, but also may lead to human errors. Especially in emergency situations, the lag and uncertainty of human judgment may lead to missing the best time to handle the fault.

[0010] Therefore, how to provide a distribution network short-circuit fault judgment method based on the load-end phase voltage amplitude change is a problem that those skilled in the art need to solve urgently. Summary of the invention

[0011] One purpose of the present invention is to propose a distribution network short-circuit fault judgment method based on the load-end phase voltage amplitude change. The present invention combines high-order statistical spectral subtraction, quantum tunneling effect detection algorithm, generalized maximum entropy principle and zero scattering theory to describe in detail the whole process from fault detection to fault classification and isolation. The method has the advantages of high detection accuracy, fast response speed, strong multi-point fault handling capability and good dynamic adaptability, and is particularly suitable for fault diagnosis and processing in complex power systems.

[0012] According to an embodiment of the present invention, a method for determining a short-circuit fault in a distribution network based on a load-terminal phase voltage amplitude variation comprises the following steps:

[0013] S1. Install voltage monitoring devices at multiple load ends of the distribution network, collect phase voltage data of each load end in real time, and generate voltage amplitude changes as input data;

[0014] S2, preprocessing the collected voltage amplitude variation to eliminate external noise interference and generate preprocessed voltage variation data;

[0015] S3. Analyze the pre-processed voltage variation data based on high-order statistical spectrum subtraction, focus on identifying abnormal components in the signal, and highlight abnormal changes by subtracting the background noise spectrum from the signal. When significant spectrum differences are detected, preliminarily judge potential short-circuit faults.

[0016] S4. Further analyze the voltage change data of potential short-circuit faults, use quantum tunneling effect detection algorithm to simulate the quantum behavior of voltage change, and confirm the abnormality when the voltage change exceeds the preset tunneling barrier, and record the abnormal location;

[0017] S5. Further time series analysis is performed on the voltage change data at the abnormal position. The instantaneous entropy value and the trend entropy value are calculated using the generalized maximum entropy principle. By identifying the duration and instantaneous characteristics of the entropy value, when the entropy value reaches a maximum or minimum value, the occurrence of a short circuit fault is confirmed.

[0018] S6. Analyze the propagation characteristics of the voltage signal of the fault point in the distribution network based on the zero scattering theory, and further accurately locate the fault point by calculating the scattering coefficient of the signal. When the scattering coefficient significantly deviates from the normal range, determine the location of the fault point;

[0019] S7. After determining the fault point, classify the voltage change characteristics of the fault point based on the minimum entropy production rate principle, determine the fault type by calculating the change pattern of the entropy production rate, and classify the fault type into single-phase, two-phase or three-phase short circuit;

[0020] S8. Fault isolation based on generalized predictive control predicts the state of the distribution network after a fault occurs, calculates the optimal isolation path and operation steps in advance, dynamically adjusts the fault isolation plan, quickly removes the fault area and restores power supply to the non-fault area.

[0021] Optionally, S3 includes the following steps:

[0022] S31, performing complex harmonic analysis on the pre-processed voltage variation data, converting the time domain signal into a complex frequency domain signal, obtaining the amplitude spectrum and phase spectrum of the signal, revealing the relative positions and change trends between the frequency components, and thus capturing the characteristics of the abnormal signal;

[0023] S32. Apply multi-window high-order statistics analysis technology to identify nonlinear characteristics and abnormal patterns of different frequency components, select multiple window functions suitable for power signal analysis, calculate high-order statistics of signals in different windows, including skewness and kurtosis, and optimize the detection sensitivity of abnormal signals by comprehensively analyzing frequency characteristics at different scales;

[0024] S33. Based on the multi-window high-order statistics analysis results, high-order statistics spectrum subtraction is performed. First, the high-order cumulants of the complex frequency domain signal are calculated to obtain the high-order statistical features of the signal. Then, the high-order cumulants spectrum of the background noise is subtracted from the high-order cumulants spectrum of the signal to extract the abnormal signal features:

[0025] ;

[0026] in, represents the characteristic spectrum of abnormal signal after high-order statistical spectrum subtraction, For the signal at time The high-order cumulant spectrum of For background noise at time The high-order cumulant spectrum of For the moment The weight coefficient is used to adjust the contribution of different time points to the abnormal characteristic spectrum. is the regularization coefficient, is the attenuation coefficient, represents the high-order cumulant spectrum, where and is the order of the cumulant, is the frequency, is the mean of the high-order cumulant spectrum;

[0027] S34. Analyze the results of the high-order statistical spectrum subtraction, and preliminarily judge the potential short-circuit fault in the signal by detecting the significant abnormal components in the high-order cumulative spectrum. Mark these abnormal data and use them as the basis for subsequent further analysis and fault location.

[0028] Optionally, S4 includes the following steps:

[0029] S41, obtaining the characteristic spectrum of the abnormal signal after high-order statistical spectrum subtraction analysis, using it as input data of the multi-level quantum tunneling effect detection algorithm, and setting multiple preset tunneling barriers according to the amplitude of the voltage change , , , corresponding to different voltage change thresholds, used to evaluate different degrees of abnormality;

[0030] S42, using a multi-level quantum tunneling effect detection algorithm, first calculate the quantum state wave function of the voltage change signal, and then determine whether the wave function can penetrate the first preset tunneling barrier , if the wave function penetrates the first barrier , then record the voltage change point at this time as the "primary abnormal" point, and mark the position and time of the abnormal point;

[0031] S43. Use the marked “primary anomaly” point as input and continue to apply the quantum tunneling effect detection algorithm to detect whether the wave function can penetrate the second preset tunneling barrier. , if the wave function penetrates the second barrier , then mark the voltage change point as a "medium abnormal" point and record its position and time;

[0032] S44. Use the marked "intermediate anomaly" point as input and continue to apply the quantum tunneling effect detection algorithm to determine whether the wave function can penetrate the third preset tunneling barrier. , if the wave function penetrates the third barrier , then mark the voltage change point as a "high-level abnormal" point and record its location and time;

[0033] S45. Calculate the comprehensive tunneling probability for the marked "primary anomaly", "intermediate anomaly" and "advanced anomaly" points :

[0034] ;

[0035] in, It represents the short-circuit fault probability after comprehensive judgment of multi-level detection, is the tunneling probability corresponding to each level of detection , , , corresponding to the "primary anomaly", "intermediate anomaly" and "advanced anomaly" points respectively, The characteristic spectrum of the abnormal signal calculated by high-order statistical spectrum subtraction corresponds to different frequencies. The following features, is the weight coefficient of different detection levels, is the regularization coefficient, is the attenuation coefficient, which is used to control the comprehensive tunneling probability and the average tunneling probability The exponential decay of is the average tunneling probability among all detection levels;

[0036] S46, the calculated comprehensive tunneling probability The fault level of each abnormal point is confirmed according to the size of the comprehensive tunneling probability by comparing it with the preset threshold, and the location and time information of the abnormal point is marked according to the fault level and stored in the fault record database of the distribution network for subsequent fault isolation and analysis.

[0037] Optionally, S5 includes the following steps:

[0038] S51, obtaining abnormal position data marked by the quantum tunneling effect detection algorithm, using these data as input, performing time series analysis, extracting the instantaneous value of the voltage change and the rate of change at adjacent moments, constructing a time series model of the voltage change, and recording the distribution of the voltage change at different spatial positions;

[0039] S52. Use the generalized maximum entropy principle to perform a spatiotemporal joint analysis on the time series of voltage changes and calculate the joint entropy value of the voltage signal in the time and space dimensions. :

[0040] ;

[0041] in, and The time and spatial location Previous The probability distribution of states, is the characteristic spectrum of abnormal signal by high-order statistics spectrum subtraction, is the comprehensive tunneling probability, and is the new adjustment parameter, is the newly introduced adjustment coefficient, is the phase angle associated with the space-time position;

[0042] S53. Based on the calculation of the spatiotemporal joint entropy value, further calculate the trend entropy value , capturing the long-term behavior pattern of voltage variation in time and space dimensions:

[0043] ;

[0044] in, For time and spatial location Previous The probability distribution of the trend, is the characteristic spectrum of abnormal signal by high-order statistics spectrum subtraction, is the comprehensive tunneling probability, and is the adjustment parameter of generalized maximum entropy;

[0045] S54. Identify the spatiotemporal joint entropy value and trend entropy The maximum and minimum positions of the fault points are determined as potential short-circuit fault points, and the time and space positions of the fault points are marked and recorded in the fault database;

[0046] S55. Classify the marked fault points, and divide the fault points into different types of faults according to the characteristics of their spatiotemporal joint entropy values ​​and trend entropy values. According to the duration and spatial distribution of the entropy values ​​in the spatiotemporal dimension, classify the faults as "local short-term anomalies" or "wide-area long-term anomalies", respectively. Generate a corresponding fault report based on the classification results and store it in the fault database of the distribution network for subsequent fault analysis and processing.

[0047] Optionally, the S6 includes the following steps:

[0048] S61, obtaining the marked fault point data, taking it as input, analyzing the propagation characteristics of the voltage signal of the fault point in the distribution network based on the zero scattering theory, constructing a propagation model of the voltage signal in the distribution network, and determining the propagation path of the voltage signal at different spatial positions;

[0049] S62. Calculate the scattering coefficient of the voltage signal during propagation :

[0050] ;

[0051] in, Fault point The scattering coefficient at is the voltage change at the fault point, represents the rate of change of the signal in space, is the voltage signal at position The propagation path of is the characteristic spectrum of abnormal signal by high-order statistics spectrum subtraction, represents the spatial curvature of the signal propagation path, is the comprehensive tunneling probability, is the space-time joint entropy, and are the starting point and end point of the propagation path, is the weight coefficient for adjusting the second-order derivative term;

[0052] S63, using multi-point scattering collaborative analysis technology, collaborative calculation of scattering coefficients for multiple fault points at the same time, by building a collaborative matrix , combined with the scattering coefficient of each fault point , the calculation formula of the synergy matrix is:

[0053] ;

[0054] in, and Fault Point and The scattering coefficient at Fault point and The phase difference between The number of fault points involved in collaborative analysis;

[0055] S64, Synergy Matrix Perform analysis to identify the interaction characteristics between fault points. When the cooperative scattering coefficients of some fault points significantly deviate from the normal range, these locations are confirmed as high-risk fault areas and their location and time information are marked.

[0056] S65. Based on the results of collaborative analysis, the fault point is further accurately located. Through multi-point scattering collaborative analysis, the relative positions and propagation paths of multiple fault points are determined. According to the degree of deviation of the collaborative scattering coefficient, an accurate positioning report of the fault point is generated, and the position of the fault point is stored in the fault record database of the distribution network for subsequent fault isolation and processing.

[0057] Optionally, the S7 includes the following steps:

[0058] S71. Obtain the determined fault point and related voltage change characteristic data, use them as input, analyze the voltage change characteristics of the fault point based on the minimum entropy productivity principle, extract the voltage change rate and acceleration of the fault point, and construct a voltage change characteristic matrix;

[0059] S72. Calculate the minimum entropy production rate at the fault point :

[0060] ;

[0061] in, The fault point at time The minimum entropy production rate on is the voltage variation characteristic of the fault point The probability distribution of The reference voltage variation characteristic of the fault point The probability distribution of is the number of voltage variation characteristics;

[0062] S73, a multi-point collaborative analysis of minimum entropy productivity is performed for multiple fault points at the same time, and the minimum entropy productivity collaborative matrix is ​​obtained by constructing an entropy productivity collaborative matrix and combining the minimum entropy productivity of each fault point:

[0063] ;

[0064] in, For the The minimum entropy production rate of the fault point, For the Minimum entropy production rate for each fault point Fault point and The phase difference between and Fault point and The scattering coefficient, The number of fault points involved in collaborative analysis;

[0065] S74. According to the analysis results of the minimum entropy production rate and the synergy matrix, determine the fault type and classify the fault type into single-phase short circuit, two-phase short circuit or three-phase short circuit. If and The result shows that a fault point is independently prominent and is determined to be a single-phase short circuit. It shows that the minimum entropy production rates of the two fault points interact significantly, and it is determined to be a two-phase short circuit. It shows that the minimum entropy production rates of multiple fault points interact strongly with each other, and the minimum entropy production rates change dramatically, which is determined to be a three-phase short circuit;

[0066] S75. Associate the classification result with the temporal and spatial location of the fault point, generate a fault classification report, and store the classification result in a fault record database of the distribution network for subsequent analysis and processing.

[0067] The beneficial effects of the present invention are:

[0068] (1) The present invention significantly improves the accuracy and speed of short-circuit fault detection and location in distribution networks by combining high-order statistical spectral subtraction, quantum tunneling effect detection algorithm, generalized maximum entropy principle and zero scattering theory. In particular, when dealing with multi-point faults in complex power grid environments, the present invention overcomes the shortcomings of traditional methods in terms of detection accuracy and response speed, can accurately identify and locate fault points, and provide real-time fault classification and processing solutions.

[0069] (2) The present invention optimizes the classification and judgment of the voltage change characteristics of the fault point by introducing the minimum entropy productivity principle, effectively improving the accuracy of fault type identification. At the same time, based on the fault isolation method of generalized predictive control, the present invention can dynamically adjust the isolation scheme and quickly restore power supply to non-fault areas, greatly reducing the time and scope of power outages.

[0070] (3) The present invention uses a variety of advanced signal processing and analysis technologies to provide an efficient, accurate and intelligent judgment and processing solution for short-circuit faults in complex power systems. It has the ability to dynamically adapt to complex power grid environments and significantly improves the comprehensiveness and real-time nature of fault detection, thereby ensuring the safe and stable operation of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0072] Figure 1 It is a flow chart of the method for judging short-circuit fault of distribution network based on the change of phase voltage amplitude at load end proposed by the present invention;

[0073] Figure 2 This is a schematic diagram of the method for identifying abnormal signal components based on high-order statistics spectral subtraction proposed by the present invention;

[0074] Figure 3 Schematic diagram of the quantum tunneling effect detection algorithm proposed in the present invention confirming and recording the fault location. DETAILED DESCRIPTION

[0075] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0076] refer to Figure 1-3 The method for judging short-circuit fault of distribution network based on the change of phase voltage amplitude at load end comprises the following steps:

[0077] S1. Install voltage monitoring devices at multiple load ends of the distribution network, collect phase voltage data of each load end in real time, and generate voltage amplitude changes as input data;

[0078] S2, preprocessing the collected voltage amplitude variation to eliminate external noise interference and generate preprocessed voltage variation data;

[0079] S3. Analyze the pre-processed voltage variation data based on high-order statistical spectrum subtraction, focus on identifying abnormal components in the signal, and highlight abnormal changes by subtracting the background noise spectrum from the signal. When significant spectrum differences are detected, preliminarily judge potential short-circuit faults.

[0080] S4. Further analyze the voltage change data of potential short-circuit faults, use quantum tunneling effect detection algorithm to simulate the quantum behavior of voltage change, and confirm the abnormality when the voltage change exceeds the preset tunneling barrier, and record the abnormal location;

[0081] S5. Further time series analysis is performed on the voltage change data at the abnormal position. The instantaneous entropy value and the trend entropy value are calculated using the generalized maximum entropy principle. By identifying the duration and instantaneous characteristics of the entropy value, when the entropy value reaches a maximum or minimum value, the occurrence of a short circuit fault is confirmed.

[0082] S6. Analyze the propagation characteristics of the voltage signal of the fault point in the distribution network based on the zero scattering theory, and further accurately locate the fault point by calculating the scattering coefficient of the signal. When the scattering coefficient significantly deviates from the normal range, determine the location of the fault point;

[0083] S7. After determining the fault point, classify the voltage change characteristics of the fault point based on the minimum entropy production rate principle, determine the fault type by calculating the change pattern of the entropy production rate, and classify the fault type into single-phase, two-phase or three-phase short circuit;

[0084] S8. Fault isolation based on generalized predictive control predicts the state of the distribution network after a fault occurs, calculates the optimal isolation path and operation steps in advance, dynamically adjusts the fault isolation plan, quickly removes the fault area and restores power supply to the non-fault area.

[0085] In this implementation, S3 includes the following steps:

[0086] S31, performing complex harmonic analysis on the pre-processed voltage variation data, converting the time domain signal into a complex frequency domain signal, obtaining the amplitude spectrum and phase spectrum of the signal, revealing the relative positions and change trends between the frequency components, and thus capturing the characteristics of the abnormal signal;

[0087] S32. Apply multi-window high-order statistics analysis technology to identify nonlinear characteristics and abnormal patterns of different frequency components, select multiple window functions suitable for power signal analysis, calculate high-order statistics of signals in different windows, including skewness and kurtosis, and optimize the detection sensitivity of abnormal signals by comprehensively analyzing frequency characteristics at different scales;

[0088] S33. Based on the multi-window high-order statistics analysis results, high-order statistics spectrum subtraction is performed. First, the high-order cumulants of the complex frequency domain signal are calculated to obtain the high-order statistical features of the signal. Then, the high-order cumulants spectrum of the background noise is subtracted from the high-order cumulants spectrum of the signal to extract the abnormal signal features:

[0089] ;

[0090] in, represents the characteristic spectrum of abnormal signal after high-order statistical spectrum subtraction, For the signal at time The high-order cumulant spectrum of For background noise at time The high-order cumulant spectrum of For the moment The weight coefficient is used to adjust the contribution of different time points to the abnormal characteristic spectrum. is the regularization coefficient, is the attenuation coefficient, represents the high-order cumulant spectrum, where and is the order of the cumulant, is the frequency, is the mean of the high-order cumulant spectrum;

[0091] S34. Analyze the results of the high-order statistical spectrum subtraction, and preliminarily judge the potential short-circuit fault in the signal by detecting the significant abnormal components in the high-order cumulative spectrum. Mark these abnormal data and use them as the basis for subsequent further analysis and fault location.

[0092] In this implementation manner, S4 includes the following steps:

[0093] S41, obtaining the characteristic spectrum of the abnormal signal after high-order statistical spectrum subtraction analysis, using it as input data of the multi-level quantum tunneling effect detection algorithm, and setting multiple preset tunneling barriers according to the amplitude of the voltage change , , , corresponding to different voltage change thresholds, used to evaluate different degrees of abnormality;

[0094] S42, using a multi-level quantum tunneling effect detection algorithm, first calculate the quantum state wave function of the voltage change signal, and then determine whether the wave function can penetrate the first preset tunneling barrier , if the wave function penetrates the first barrier , then record the voltage change point at this time as the "primary abnormal" point, and mark the position and time of the abnormal point;

[0095] S43. Use the marked “primary anomaly” point as input and continue to apply the quantum tunneling effect detection algorithm to detect whether the wave function can penetrate the second preset tunneling barrier. , if the wave function penetrates the second barrier , then mark the voltage change point as a "medium abnormal" point and record its position and time;

[0096] S44. Use the marked "intermediate anomaly" point as input and continue to apply the quantum tunneling effect detection algorithm to determine whether the wave function can penetrate the third preset tunneling barrier. , if the wave function penetrates the third barrier , then mark the voltage change point as a "high-level abnormal" point and record its location and time;

[0097] S45. Calculate the comprehensive tunneling probability for the marked "primary anomaly", "intermediate anomaly" and "advanced anomaly" points :

[0098] ;

[0099] in, It represents the short-circuit fault probability after comprehensive judgment of multi-level detection, is the tunneling probability corresponding to each level of detection , , , corresponding to the "primary anomaly", "intermediate anomaly" and "advanced anomaly" points respectively, The characteristic spectrum of the abnormal signal calculated by high-order statistical spectrum subtraction corresponds to different frequencies. The following features, is the weight coefficient of different detection levels, is the regularization coefficient, is the attenuation coefficient, which is used to control the comprehensive tunneling probability and the average tunneling probability The exponential decay of is the average tunneling probability among all detection levels;

[0100] S46, the calculated comprehensive tunneling probability The fault level of each abnormal point is confirmed according to the size of the comprehensive tunneling probability by comparing it with the preset threshold, and the location and time information of the abnormal point is marked according to the fault level and stored in the fault record database of the distribution network for subsequent fault isolation and analysis.

[0101] In this implementation manner, S5 includes the following steps:

[0102] S51, obtaining abnormal position data marked by the quantum tunneling effect detection algorithm, using these data as input, performing time series analysis, extracting the instantaneous value of the voltage change and the rate of change at adjacent moments, constructing a time series model of the voltage change, and recording the distribution of the voltage change at different spatial positions;

[0103] S52. Use the generalized maximum entropy principle to perform a spatiotemporal joint analysis on the time series of voltage changes and calculate the joint entropy value of the voltage signal in the time and space dimensions. :

[0104] ;

[0105] in, and The time and spatial location Previous The probability distribution of states, is the characteristic spectrum of abnormal signal by high-order statistics spectrum subtraction, is the comprehensive tunneling probability, and is the new adjustment parameter, is the newly introduced adjustment coefficient, is the phase angle associated with the space-time position;

[0106] S53. Based on the calculation of the spatiotemporal joint entropy value, further calculate the trend entropy value , capturing the long-term behavior pattern of voltage variation in time and space dimensions:

[0107] ;

[0108] in, For time and spatial location Previous The probability distribution of the trend, is the characteristic spectrum of abnormal signal by high-order statistics spectrum subtraction, is the comprehensive tunneling probability, and is the adjustment parameter of generalized maximum entropy;

[0109] S54. Identify the spatiotemporal joint entropy value and trend entropy The maximum and minimum positions of the fault points are determined as potential short-circuit fault points, and the time and space positions of the fault points are marked and recorded in the fault database;

[0110] S55. Classify the marked fault points, and divide the fault points into different types of faults according to the characteristics of their spatiotemporal joint entropy values ​​and trend entropy values. According to the duration and spatial distribution of the entropy values ​​in the spatiotemporal dimension, classify the faults as "local short-term anomalies" or "wide-area long-term anomalies", respectively. Generate a corresponding fault report based on the classification results and store it in the fault database of the distribution network for subsequent fault analysis and processing.

[0111] In this implementation manner, S6 includes the following steps:

[0112] S61, obtaining the marked fault point data, taking it as input, analyzing the propagation characteristics of the voltage signal of the fault point in the distribution network based on the zero scattering theory, constructing a propagation model of the voltage signal in the distribution network, and determining the propagation path of the voltage signal at different spatial positions;

[0113] S62. Calculate the scattering coefficient of the voltage signal during propagation :

[0114] ;

[0115] in, Fault point The scattering coefficient at is the voltage change at the fault point, represents the rate of change of the signal in space, is the voltage signal at position The propagation path of is the characteristic spectrum of abnormal signal by high-order statistics spectrum subtraction, represents the spatial curvature of the signal propagation path, is the comprehensive tunneling probability, is the space-time joint entropy, and are the starting point and end point of the propagation path, is the weight coefficient for adjusting the second-order derivative term;

[0116] S63, using multi-point scattering collaborative analysis technology, collaborative calculation of scattering coefficients for multiple fault points at the same time, by building a collaborative matrix , combined with the scattering coefficient of each fault point , the calculation formula of the synergy matrix is:

[0117] ;

[0118] in, and Fault Point and The scattering coefficient at Fault point and The phase difference between The number of fault points involved in collaborative analysis;

[0119] S64, Synergy Matrix Perform analysis to identify the interaction characteristics between fault points. When the cooperative scattering coefficients of some fault points significantly deviate from the normal range, these locations are confirmed as high-risk fault areas and their location and time information are marked.

[0120] S65. Based on the results of collaborative analysis, the fault point is further accurately located. Through multi-point scattering collaborative analysis, the relative positions and propagation paths of multiple fault points are determined. According to the degree of deviation of the collaborative scattering coefficient, an accurate positioning report of the fault point is generated, and the position of the fault point is stored in the fault record database of the distribution network for subsequent fault isolation and processing.

[0121] In this implementation manner, the S7 includes the following steps:

[0122] S71. Obtain the determined fault point and related voltage change characteristic data, use them as input, analyze the voltage change characteristics of the fault point based on the minimum entropy productivity principle, extract the voltage change rate and acceleration of the fault point, and construct a voltage change characteristic matrix;

[0123] S72. Calculate the minimum entropy production rate at the fault point :

[0124] ;

[0125] in, The fault point at time The minimum entropy production rate on is the voltage variation characteristic of the fault point The probability distribution of The reference voltage variation characteristic of the fault point The probability distribution of is the number of voltage variation characteristics;

[0126] S73, a multi-point collaborative analysis of minimum entropy productivity is performed for multiple fault points at the same time, and the minimum entropy productivity collaborative matrix is ​​obtained by constructing an entropy productivity collaborative matrix and combining the minimum entropy productivity of each fault point:

[0127] ;

[0128] in, For the The minimum entropy production rate of the fault point, For the Minimum entropy production rate for each fault point Fault point and The phase difference between and Fault point and The scattering coefficient, The number of fault points involved in collaborative analysis;

[0129] S74. According to the analysis results of the minimum entropy production rate and the synergy matrix, determine the fault type and classify the fault type into single-phase short circuit, two-phase short circuit or three-phase short circuit. If and The result shows that a fault point is independently prominent and is determined to be a single-phase short circuit. It shows that the minimum entropy production rates of the two fault points interact significantly, and it is determined to be a two-phase short circuit. It shows that the minimum entropy production rates of multiple fault points interact strongly with each other, and the minimum entropy production rates change dramatically, which is determined to be a three-phase short circuit;

[0130] S75. Associate the classification result with the temporal and spatial location of the fault point, generate a fault classification report, and store the classification result in a fault record database of the distribution network for subsequent analysis and processing.

[0131] Embodiment 1:

[0132] In order to verify the feasibility of the present invention in implementation, the present invention is applied to the distribution network of a large city in China to detect and handle short-circuit faults during high-load periods. The city's distribution network covers multiple substations and hundreds of load terminals. Due to high power demand and complex lines, short-circuit faults occur from time to time, seriously affecting the normal operation of the city. Faced with this situation, the power sector is in urgent need of a technology that can quickly and accurately locate the fault point and handle it in a timely manner. In this embodiment, we selected the distribution network short-circuit fault judgment method based on the load-end phase voltage amplitude change of the present invention for fault detection and processing.

[0133] During the implementation process, we first installed voltage monitoring devices at multiple load ends of the distribution network. These devices can collect phase voltage data in real time and convert it into voltage amplitude changes as input data. Then, these data are transmitted to the central monitoring system for processing. The central monitoring system pre-processes the collected voltage amplitude change data, eliminates external noise interference using advanced filtering technology, and generates pre-processed voltage change data.

[0134] Table 1: Voltage monitoring and processing

[0135]

[0136] After preprocessing, the data is passed to the high-order statistics spectral subtraction module. This module performs detailed signal processing on the data, including high-order feature extraction and calculation of high-order cumulants. Through spectral subtraction, the system successfully identified the abnormal components in the signal. Preliminary analysis shows that the detected abnormal signal has the characteristics of a short-circuit fault. In order to further confirm the fault point, the system starts the quantum tunneling effect detection algorithm. By simulating the quantum behavior of voltage changes, the system sets multiple preset tunneling barriers and determines whether the wave function can penetrate these barriers. When the voltage change exceeds the preset tunneling barrier, the system confirms the location of multiple abnormal points and accurately records the time and spatial positions of these abnormal points.

[0137] Next, the system used the generalized maximum entropy principle to perform a time series analysis on the voltage change data at the fault point. By calculating the instantaneous entropy value and the trend entropy value, the system identified the duration and instantaneous characteristics of the entropy value, and further confirmed the occurrence of the short circuit fault. The application of the generalized maximum entropy principle enables the system to identify the key characteristics of the fault in complex time series data. Subsequently, the system combined the zero scattering theory to conduct a detailed analysis of the voltage signal propagation characteristics of the fault point and accurately located the fault point.

[0138] Table 2: Fault detection and isolation effects

[0139]

[0140] After locating the fault point, the system classified the voltage change characteristics of the fault point according to the principle of minimum entropy production rate. By calculating the change pattern of entropy production rate, the system determined that the fault was a single-phase short circuit. In order to quickly restore power supply, the system started the fault isolation method based on generalized predictive control, dynamically adjusted the isolation scheme, quickly removed the fault area, and restored power supply to the non-fault area.

[0141] Table 3: Statistics after troubleshooting

[0142]

[0143] The system performed very well during the fault detection and handling process. The entire process from fault detection to fault isolation took only 1.5 seconds, while traditional methods usually take 5 to 10 seconds. This efficient response time significantly reduced the impact of the fault on the city's power supply. Data analysis showed that at key nodes, the voltage fluctuation amplitude reached 20%, and through the quantum tunneling effect detection algorithm, the system successfully located 3 major abnormal points, and these abnormal points were all actual fault points. In addition, the system's fault isolation time was 0.8 seconds, ensuring that 80% of the load ends restored power supply within 5 seconds after the fault occurred, greatly reducing the duration and scope of power outages.

[0144] After the fault handling was completed, the system automatically generated a detailed fault diagnosis report. The report showed that the short-circuit fault involved a line length of 12 kilometers and affected 12,500 users. The fault detection and isolation method of the present invention avoided long-term power outages for about 8,000 users, ensuring the continuous power supply of key urban infrastructure. This application not only verified the practicality and efficiency of the present invention, but also demonstrated its wide applicability in complex power systems, significantly improving the stability and safety of the distribution network.

[0145] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for judging short-circuit faults in a distribution network based on the change in load-end phase voltage amplitude, characterized in that: The steps include: S1. Install voltage monitoring devices at multiple load ends of the distribution network, collect phase voltage data of each load end in real time, and generate voltage amplitude changes as input data; S2, preprocessing the collected voltage amplitude variation to eliminate external noise interference and generate preprocessed voltage variation data; S3. Analyze the pre-processed voltage variation data based on high-order statistical spectrum subtraction, focus on identifying abnormal components in the signal, and highlight abnormal changes by subtracting the background noise spectrum from the signal. When significant spectrum differences are detected, preliminarily judge potential short-circuit faults. S4. Further analyze the voltage change data of potential short-circuit faults, use quantum tunneling effect detection algorithm to simulate the quantum behavior of voltage change, and confirm the abnormality when the voltage change exceeds the preset tunneling barrier, and record the abnormal location; S5. Further time series analysis is performed on the voltage change data at the abnormal position. The instantaneous entropy value and the trend entropy value are calculated using the generalized maximum entropy principle. By identifying the duration and instantaneous characteristics of the entropy value, when the entropy value reaches a maximum or minimum value, the occurrence of a short circuit fault is confirmed. S6. Analyze the propagation characteristics of the voltage signal of the fault point in the distribution network based on the zero scattering theory, and further accurately locate the fault point by calculating the scattering coefficient of the signal. When the scattering coefficient significantly deviates from the normal range, determine the location of the fault point; S7. After determining the fault point, classify the voltage change characteristics of the fault point based on the minimum entropy production rate principle, determine the fault type by calculating the change pattern of the entropy production rate, and classify the fault type into single-phase, two-phase or three-phase short circuit; S8. Fault isolation based on generalized predictive control: by predicting the state of the distribution network after a fault occurs, the optimal isolation path and operation steps are calculated in advance, the fault isolation plan is dynamically adjusted, the fault area is quickly removed, and the power supply to the non-fault area is restored; The S3 specifically includes: S31, performing complex harmonic analysis on the pre-processed voltage variation data, converting the time domain signal into a complex frequency domain signal, obtaining the amplitude spectrum and phase spectrum of the signal, revealing the relative positions and change trends between the frequency components, and thus capturing the characteristics of the abnormal signal; S32. Apply multi-window high-order statistics analysis technology to identify nonlinear characteristics and abnormal patterns of different frequency components, select multiple window functions suitable for power signal analysis, calculate high-order statistics of signals in different windows, including skewness and kurtosis, and optimize the detection sensitivity of abnormal signals by comprehensively analyzing frequency characteristics at different scales; S33. Based on the multi-window high-order statistics analysis results, high-order statistics spectrum subtraction is performed. First, the high-order cumulants of the complex frequency domain signal are calculated to obtain the high-order statistical features of the signal. Then, the high-order cumulants spectrum of the background noise is subtracted from the high-order cumulants spectrum of the signal to extract the abnormal signal features: ; in, represents the characteristic spectrum of abnormal signal after high-order statistical spectrum subtraction, For the signal at time The high-order cumulant spectrum of For background noise at time The high-order cumulant spectrum of For the moment The weight coefficient is used to adjust the contribution of different time points to the abnormal characteristic spectrum. is the regularization coefficient, is the attenuation coefficient, represents the high-order cumulant spectrum, where and is the order of the cumulant, is the frequency, is the mean of the high-order cumulant spectrum; S34, analyzing the results of the high-order statistical spectrum subtraction, preliminarily judging the potential short-circuit fault in the signal by detecting the significant abnormal components in the high-order cumulative spectrum, marking these abnormal data, and using them as the basis for subsequent further analysis and fault location; The S4 specifically includes: S41, obtaining the characteristic spectrum of the abnormal signal after high-order statistical spectrum subtraction analysis, using it as input data of the multi-level quantum tunneling effect detection algorithm, and setting multiple preset tunneling barriers according to the amplitude of the voltage change , , , corresponding to different voltage change thresholds, used to evaluate different degrees of abnormality; S42, using a multi-level quantum tunneling effect detection algorithm, first calculate the quantum state wave function of the voltage change signal, and then determine whether the wave function can penetrate the first preset tunneling barrier , if the wave function penetrates the first barrier , then record the voltage change point at this time as the "primary abnormal" point, and mark the position and time of the abnormal point; S43, using the marked "primary anomaly" point as input, continue to apply the quantum tunneling effect detection algorithm to detect whether the wave function can penetrate the second preset tunneling barrier , if the wave function penetrates the second barrier , then mark the voltage change point as a "medium abnormal" point and record its position and time; S44. Use the marked "intermediate anomaly" point as input and continue to apply the quantum tunneling effect detection algorithm to determine whether the wave function can penetrate the third preset tunneling barrier. , if the wave function penetrates the third barrier , then mark the voltage change point as a "high-level abnormal" point and record its location and time; S45. Calculate the comprehensive tunneling probability for the marked "primary anomaly", "intermediate anomaly" and "advanced anomaly" points : ； in, It represents the short-circuit fault probability after comprehensive judgment of multi-level detection, is the tunneling probability corresponding to each level of detection , , , corresponding to the "primary anomaly", "intermediate anomaly" and "advanced anomaly" points respectively, The characteristic spectrum of the abnormal signal calculated by high-order statistical spectrum subtraction corresponds to different frequencies. The following features, is the weight coefficient of different detection levels, is the regularization coefficient, is the attenuation coefficient, which is used to control the comprehensive tunneling probability and the average tunneling probability The exponential decay of is the average tunneling probability among all detection levels; S46, the calculated comprehensive tunneling probability Compare with the preset threshold, confirm the fault level of each abnormal point according to the size of the comprehensive tunneling probability, and mark the location and time information of the abnormal point according to the fault level, and store it in the fault record database of the distribution network for subsequent fault isolation and analysis; The S5 specifically includes: S51, obtaining abnormal position data marked by the quantum tunneling effect detection algorithm, using these data as input, performing time series analysis, extracting the instantaneous value of the voltage change and the rate of change at adjacent moments, constructing a time series model of the voltage change, and recording the distribution of the voltage change at different spatial positions; S52. Use the generalized maximum entropy principle to perform a spatiotemporal joint analysis on the time series of voltage changes and calculate the joint entropy value of the voltage signal in the time and space dimensions. : ； in, and The time and spatial location Previous The probability distribution of states, is the characteristic spectrum of abnormal signal by high-order statistics spectrum subtraction, is the comprehensive tunneling probability, and is the new adjustment parameter, is the newly introduced adjustment coefficient, is the phase angle associated with the space-time position; S53. Based on the calculation of the spatiotemporal joint entropy value, further calculate the trend entropy value , capturing the long-term behavior pattern of voltage variation in time and space dimensions: ； in, For time and spatial location Previous The probability distribution of the trend, is the characteristic spectrum of abnormal signal by high-order statistics spectrum subtraction, is the comprehensive tunneling probability, and is the adjustment parameter of generalized maximum entropy; S54. Identify the spatiotemporal joint entropy value and trend entropy The maximum and minimum positions of the fault points are determined as potential short-circuit fault points, and the time and space positions of the fault points are marked and recorded in the fault database; S55, classifying the marked fault points, classifying the fault points into different types of faults according to the characteristics of their spatiotemporal joint entropy values ​​and trend entropy values, and classifying the faults into "local short-term anomalies" or "wide-area long-term anomalies" according to the duration and spatial distribution of the entropy values ​​in the spatiotemporal dimension, and generating corresponding fault reports according to the classification results, and storing them in the fault database of the distribution network for subsequent fault analysis and processing; The S6 specifically includes: S61, obtaining the marked fault point data, taking it as input, analyzing the propagation characteristics of the voltage signal of the fault point in the distribution network based on the zero scattering theory, constructing a propagation model of the voltage signal in the distribution network, and determining the propagation path of the voltage signal at different spatial positions; S62. Calculate the scattering coefficient of the voltage signal during propagation : ； in, Fault point The scattering coefficient at is the voltage change at the fault point, represents the rate of change of the signal in space, is the voltage signal at position The propagation path of is the characteristic spectrum of abnormal signal by high-order statistics spectrum subtraction, represents the spatial curvature of the signal propagation path, is the comprehensive tunneling probability, is the space-time joint entropy, and are the starting and ending points of the propagation path, is the weight coefficient for adjusting the second-order derivative term; S63, using multi-point scattering collaborative analysis technology, collaborative calculation of scattering coefficients for multiple fault points at the same time, by building a collaborative matrix , combined with the scattering coefficient of each fault point , the calculation formula of the synergy matrix is: ； in, and Fault Point and The scattering coefficient at Fault point and The phase difference between The number of fault points involved in collaborative analysis; S64, Synergy Matrix Perform analysis to identify the interaction characteristics between fault points. When the cooperative scattering coefficients of some fault points significantly deviate from the normal range, these locations are confirmed as high-risk fault areas and their location and time information are marked. S65. Based on the result of the collaborative analysis, the fault point is further accurately located. The relative positions and propagation paths of multiple fault points are determined through multi-point scattering collaborative analysis. According to the degree of deviation of the collaborative scattering coefficient, an accurate positioning report of the fault point is generated, and the position of the fault point is stored in the fault record database of the distribution network for subsequent fault isolation and processing. The S7 specifically includes: S71. Obtain the determined fault point and related voltage change characteristic data, use them as input, analyze the voltage change characteristics of the fault point based on the minimum entropy productivity principle, extract the voltage change rate and acceleration of the fault point, and construct a voltage change characteristic matrix; S72. Calculate the minimum entropy production rate at the fault point : ； in, The fault point at time The minimum entropy production rate on is the voltage variation characteristic of the fault point The probability distribution of The reference voltage variation characteristic of the fault point The probability distribution of is the number of voltage variation characteristics; S73, a multi-point collaborative analysis of minimum entropy productivity is performed for multiple fault points at the same time, and the minimum entropy productivity collaborative matrix is ​​obtained by constructing an entropy productivity collaborative matrix and combining the minimum entropy productivity of each fault point: ； in, For the The minimum entropy production rate of the fault point, For the Minimum entropy production rate for each fault point Fault point and The phase difference between and Fault point and The scattering coefficient, The number of fault points involved in collaborative analysis; S74. According to the analysis results of the minimum entropy production rate and the synergy matrix, determine the fault type and classify the fault type into single-phase short circuit, two-phase short circuit or three-phase short circuit. If and The result shows that a fault point is independently prominent and is determined to be a single-phase short circuit. It shows that the minimum entropy production rates of the two fault points interact significantly, and it is determined to be a two-phase short circuit. It shows that the minimum entropy production rates of multiple fault points interact strongly with each other, and the minimum entropy production rates change dramatically, which is determined to be a three-phase short circuit; S75. Associate the classification result with the temporal and spatial location of the fault point, generate a fault classification report, and store the classification result in a fault record database of the distribution network for subsequent analysis and processing.

Images