Power distribution network switch fault intelligent diagnosis method and device and electronic equipment

Abstract

The invention discloses a power distribution network switch fault intelligent diagnosis method and device and electronic equipment, and relates to the field of power systems. The method comprises the following steps: acquiring operation state data of a power distribution network switch; constructing a multi-dimensional feature template of the power distribution network switch in a normal state, wherein the multi-dimensional feature template is obtained by training historical operation state data; calculating the similarity between the operation state data and each feature template to obtain a plurality of similarity values; fusing the plurality of similarity values to obtain a comprehensive health index of the power distribution network switch; the comprehensive health index is compared with a preset fault threshold value, and whether the power distribution network switch breaks down or not is judged; if it is determined that the comprehensive health index is greater than or equal to a preset fault threshold, determining that the power distribution network switch does not have a fault; and if it is determined that the comprehensive health index is smaller than a preset fault threshold, determining that the power distribution network switch has a fault. By implementing the technical scheme provided by the invention, the power distribution network switch fault diagnosis efficiency is improved.

Description

Technical Field

[0001] The present application relates to the field of power systems, and specifically to a method, device and electronic equipment for intelligent diagnosis of distribution network switch faults. Background Art

[0002] With the rapid development and increasing intelligence of power systems, the stable and safe operation of distribution networks has become increasingly important. As key equipment in power systems, the health status of distribution network switches directly affects the reliability and power supply quality of the entire power grid.

[0003] At present, fault diagnosis in many power systems, especially in some more traditional distribution networks, still relies on manual judgment. This method is mainly based on the experience of on-site technicians and regular equipment inspections, which is time-consuming. A single inspection may take hours or even days, resulting in low efficiency in fault diagnosis.

[0004] Therefore, there is an urgent need for an intelligent diagnosis method, device and electronic equipment for distribution network switch faults. Summary of the invention

[0005] The present application provides a distribution network switch fault intelligent diagnosis method, device and electronic equipment, which improves the efficiency of distribution network switch fault diagnosis.

[0006] In a first aspect of the present application, a method for intelligent diagnosis of distribution network switch faults is provided, the method comprising: obtaining operating status data of the distribution network switch, the operating status data comprising current data, voltage data, temperature data and vibration data of the distribution network switch; constructing a multi-dimensional feature template of the distribution network switch in a normal state, the multi-dimensional feature template comprising multiple feature templates, the feature template comprising a current feature template, a voltage feature template, a temperature feature template and a vibration feature template, the multi-dimensional feature template being trained by historical operating status data; calculating the similarity between the operating status data and each of the feature templates to obtain multiple similarity values; fusing the multiple similarity values ​​to obtain a comprehensive health index of the distribution network switch; comparing the comprehensive health index with a preset fault threshold to determine whether the distribution network switch has a fault; if it is determined that the comprehensive health index is greater than or equal to the preset fault threshold, it is determined that the distribution network switch has not a fault; if it is determined that the comprehensive health index is less than the preset fault threshold, it is determined that the distribution network switch has a fault.

[0007] By adopting the above technical solution, by obtaining the operating status data of the distribution network switch, constructing a multi-dimensional feature template under normal state, calculating the similarity between the real-time operating state and the feature template, and integrating multiple similarity values ​​to obtain a comprehensive health index, the comprehensive health index is finally compared with the preset fault threshold, and the intelligent diagnosis of the distribution network switch fault is realized. This method makes full use of the multi-source heterogeneous data in the switch operation process, characterizes the health state of the switch from multiple angles such as current, voltage, temperature, and vibration, and improves the comprehensiveness and accuracy of fault diagnosis. At the same time, by constructing a feature template under normal state, it avoids dependence on a large amount of fault data, and reduces the difficulty and cost of model training. This method can monitor the operating status of the distribution network switch in real time, discover potential fault hazards in time, and provide reliable guarantee for the safe and stable operation of the distribution network. At the same time, through the quantified comprehensive health index, it provides intuitive decision-making basis for operation and maintenance personnel, and improves the intelligence level and work efficiency of distribution network operation and maintenance.

[0008] Optionally, constructing a multi-dimensional feature template of the distribution network switch in a normal state specifically includes: acquiring historical operating status data of multiple normally operating distribution network switches to obtain a historical data set, wherein the historical data set includes historical current data, historical voltage data, historical temperature data, and historical vibration data; performing a preprocessing operation on the historical data set to obtain a target historical data set; the preprocessing operation includes outlier removal, data normalization, and data smoothing; utilizing a sliding time window method to extract multi-domain features of the target historical data set at different time scales to obtain a feature sample set corresponding to each of the historical operating status data, wherein the multi-domain features include time domain features, frequency domain features, and time-frequency domain features; based on the feature sample sets corresponding to each of the historical operating status data, respectively training support vector data description models to obtain the current feature template, the voltage feature template, the temperature feature template, and the vibration feature template.

[0009] By adopting the above technical scheme, the historical operating status data of multiple normally operating distribution network switches are collected, and the high-quality target historical data set is obtained after preprocessing operations such as outlier removal, data normalization and smoothing. Then, the sliding time window method is used to extract the multi-domain features of the target historical data set at different time scales, including time domain features, frequency domain features and time-frequency domain features, to obtain a comprehensive and rich feature sample set. Finally, the support vector data description model is trained based on the feature sample set to obtain current feature templates, voltage feature templates, temperature feature templates and vibration feature templates. This method fully mines the health status information contained in the historical data, and effectively improves the representation ability and generalization ability of the feature template through multi-domain feature extraction and data preprocessing. The feature template can accurately characterize the normal operating mode of the distribution network switch and provide a reliable reference standard for subsequent fault diagnosis. At the same time, the support vector data description model has good anomaly detection ability, can effectively identify fault states that deviate from the normal mode, and improve the sensitivity and accuracy of fault diagnosis.

[0010] Optionally, the similarity between the operating status data and each feature template is calculated to obtain multiple similarity values, specifically including: determining the corresponding features of the operating status data relative to the feature template to obtain a corresponding feature sequence to be measured, the feature sequence to be measured including a current feature sequence, a voltage feature sequence, a temperature feature sequence and a vibration feature sequence; calculating the optimal matching distance between the feature sequence to be measured and the feature template; within a preset time window, calculating the average value of the optimal matching distance to obtain an average matching distance, and performing an exponential transformation on the average matching distance to obtain a similarity value.

[0011] By adopting the above technical solution, the corresponding features of the operating status data relative to the feature template are determined to obtain the feature sequence to be tested, and then the optimal matching distance between the feature sequence to be tested and the feature template is calculated, the average value of the optimal matching distance is calculated within the preset time window, and the average matching distance is exponentially transformed to obtain the similarity value. This method fully considers the time series characteristics of the operating state of the distribution network switch, and measures the difference between the feature sequence to be tested and the feature template through the optimal matching distance, so as to accurately evaluate the health state of the switch. At the same time, by averaging the optimal matching distance within the time window, the fluctuation and noise interference of a single match are effectively reduced, and the stability and reliability of the similarity calculation are improved. The exponential transformation maps the matching distance to the similarity space, so that the similarity value more intuitively represents the quality of the health state. This method can dynamically track the changes in the operating state of the distribution network switch, evaluate its health level in real time, provide a quantitative basis for fault diagnosis and predictive maintenance, and improve the pertinence and effectiveness of distribution network operation and maintenance.

[0012] Optionally, the calculation formula for calculating the optimal matching distance between the feature sequence to be tested and the feature template is: ; Where D(i, j) represents the cumulative distance between the first i elements of the feature sequence to be tested and the first j elements of the feature template, x i represents the i-th element of the feature sequence to be tested, y j represents the jth element of the feature template, (x i -y j ) 2 Represents the Euclidean distance between the i-th element of the feature sequence to be tested and the j-th element of the feature template.

[0013] By adopting the above technical solution, the optimal matching distance between the feature sequence to be tested and the feature template can be efficiently calculated through a recursive formula. This formula is based on the idea of ​​dynamic programming, makes full use of the optimal solution of the calculated sub-problem, avoids repeated calculations, and greatly improves the efficiency of distance calculation. At the same time, by comparing the cumulative distance between the current element and the three directions, the optimal matching path is obtained, which can effectively handle the expansion and offset of the sequence and improve the fault tolerance and robustness of the matching. The square of the Euclidean distance, as a local distance metric, can sensitively reflect the subtle differences in the eigenvalues ​​and improve the matching accuracy. This formula can quickly and accurately evaluate the similarity between the feature sequence to be tested and the feature template, and provide a reliable quantitative indicator for fault diagnosis. At the same time, by optimizing the matching path, the abnormal change pattern of the switch operation state can be identified, providing important clues for fault cause analysis and location, and improving the intelligent level of distribution network operation and maintenance.

[0014] Optionally, within the preset time window, the average value of the optimal matching distance is calculated to obtain a specific calculation formula for the average matching distance: ; in, is the average matching distance, n is the number of the preset time windows, is the optimal matching distance in the kth time window.

[0015] By adopting the above technical solution, the average matching distance is obtained by calculating the average value of the optimal matching distance within the preset time window. This method fully considers the continuity and time correlation of the operating state of the distribution network switch, extracts the optimal matching distance sequence by sliding the time window, and can dynamically reflect the changing trend of the health state of the switch. The average value of the optimal matching distance sequence can effectively smooth the fluctuation and noise interference of a single match, and improve the stability and reliability of the matching result. At the same time, by adjusting the size of the preset time window, the time scale of the health assessment can be flexibly controlled, taking into account both short-term and long-term operating state changes. In practical applications, the average matching distance can quantify the overall health level of the distribution network switch over a period of time, providing an important basis for fault trend prediction and health state tracking. By updating the average matching distance in real time, abnormal deviations of the switch operating state can be discovered in time, providing reliable support for fault warning and maintenance decisions, and improving the initiative and real-time performance of distribution network operation and maintenance.

[0016] Optionally, the specific calculation formula for performing exponential transformation on the average matching distance to obtain the similarity value is: ; Wherein, S is the similarity value, λ is the similarity adjustment factor, is the average matching distance.

[0017] By adopting the above technical solution, the similarity value is obtained by exponentially transforming the average matching distance. The exponential transformation can map the average matching distance from the Euclidean space to the similarity space, so that the similarity value falls between 0 and 1, and more intuitively represents the health status of the distribution network switch. Through the similarity adjustment factor, the nonlinearity of the exponential transformation can be flexibly controlled to balance the matching accuracy and robustness.

[0018] Optionally, the specific formula for fusing the multiple similarity values ​​to obtain the comprehensive health index of the distribution network switch is: ; Wherein, H is the comprehensive health index, N is the number of feature templates, and w i is the weight of the i-th feature template, S i is the similarity value of the i-th feature template, α i is the exponential coefficient for adjusting the i-th feature template.

[0019] By adopting the above technical solution, the similarity values ​​of multiple feature dimensions are fused into a comprehensive health index through the weighted exponential average formula. This formula fully considers the differences in the impact of different feature dimensions on the health status of distribution network switches. By introducing weight factors, the importance of each feature dimension is quantified, so that the comprehensive health index more accurately reflects the overall health level of the switch. At the same time, the exponential transformation maps the similarity value from linear space to exponential space. By adjusting the exponential coefficient, the nonlinear contribution of different features can be flexibly controlled, the influence of key features can be highlighted, and the interference of secondary features can be suppressed. In practical applications, the comprehensive health index provides a comprehensive switch health assessment indicator, which integrates the operating status information of multiple dimensions such as current, voltage, temperature, and vibration, greatly simplifying the judgment and decision-making process of operation and maintenance personnel.

[0020] In a second aspect of the present application, a distribution network switch fault intelligent diagnosis device is provided, the device includes an acquisition module and a processing module, wherein: the acquisition module is used to acquire the operating status data of the distribution network switch, the operating status data includes the current data, voltage data, temperature data and vibration data of the distribution network switch; the processing module is used to construct a multi-dimensional feature template of the distribution network switch in a normal state, the multi-dimensional feature template includes multiple feature templates, the feature template includes a current feature template, a voltage feature template, a temperature feature template and a vibration feature template, and the multi-dimensional feature template is obtained by training the historical operating status data; the processing module The module is also used to calculate the similarity between the operating status data and each of the feature templates to obtain multiple similarity values; the processing module is also used to fuse multiple similarity values ​​to obtain a comprehensive health index of the distribution network switch; the processing module is also used to compare the comprehensive health index with a preset fault threshold to determine whether the distribution network switch has a fault; the processing module is also used to determine that the distribution network switch has not a fault if it is determined that the comprehensive health index is greater than or equal to the preset fault threshold; the processing module is also used to determine that the distribution network switch has a fault if it is determined that the comprehensive health index is less than the preset fault threshold.

[0021] In the third aspect of the present application, an electronic device is provided, including a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes any one of the methods described above.

[0022] In a fourth aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores instructions, and when the instructions are executed, any of the methods described above is executed.

[0023] In summary, one or more technical solutions provided in this application have at least the following technical effects or advantages: 1. By acquiring the operating status data of the distribution network switch, constructing a multi-dimensional feature template under normal conditions, calculating the similarity between the real-time operating status and the feature template, and integrating multiple similarity values ​​to obtain a comprehensive health index, the comprehensive health index is finally compared with the preset fault threshold, realizing the intelligent diagnosis of distribution network switch faults. This method makes full use of multi-source heterogeneous data during the operation of the switch, characterizes the health status of the switch from multiple angles such as current, voltage, temperature, and vibration, and improves the comprehensiveness and accuracy of fault diagnosis. At the same time, by constructing a feature template under normal conditions, it avoids dependence on a large amount of fault data, and reduces the difficulty and cost of model training. This method can monitor the operating status of the distribution network switch in real time, discover potential fault hazards in a timely manner, and provide reliable protection for the safe and stable operation of the distribution network. At the same time, through the quantified comprehensive health index, it provides intuitive decision-making basis for operation and maintenance personnel, and improves the intelligence level and work efficiency of distribution network operation and maintenance.

[0024] 2. By collecting the historical operating status data of multiple normally operating distribution network switches, a high-quality target historical data set is obtained after preprocessing operations such as outlier removal, data normalization and smoothing. Then, the sliding time window method is used to extract the multi-domain features of the target historical data set at different time scales, including time domain features, frequency domain features and time-frequency domain features, to obtain a comprehensive and rich feature sample set. Finally, the support vector data description model is trained based on the feature sample set to obtain current feature templates, voltage feature templates, temperature feature templates and vibration feature templates. This method fully mines the health status information contained in the historical data, and effectively improves the representation and generalization capabilities of the feature templates through multi-domain feature extraction and data preprocessing. The feature template can accurately characterize the normal operating mode of the distribution network switch and provide a reliable reference standard for subsequent fault diagnosis. At the same time, the support vector data description model has good anomaly detection capabilities, can effectively identify fault states that deviate from the normal mode, and improve the sensitivity and accuracy of fault diagnosis.

[0025] 3. By determining the corresponding features of the operating status data relative to the feature template, the feature sequence to be tested is obtained, and then the optimal matching distance between the feature sequence to be tested and the feature template is calculated. The average value of the optimal matching distance is calculated within the preset time window, and the average matching distance is exponentially transformed to obtain the similarity value. This method fully considers the time series characteristics of the operating status of the distribution network switch, and measures the difference between the feature sequence to be tested and the feature template through the optimal matching distance, so as to accurately evaluate the health status of the switch. At the same time, by averaging the optimal matching distance within the time window, the fluctuation and noise interference of a single match are effectively reduced, and the stability and reliability of the similarity calculation are improved. The exponential transformation maps the matching distance to the similarity space, so that the similarity value more intuitively represents the health status. This method can dynamically track the changes in the operating status of the distribution network switch, evaluate its health level in real time, provide a quantitative basis for fault diagnosis and predictive maintenance, and improve the pertinence and effectiveness of distribution network operation and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a flow chart of a method for intelligent diagnosis of switch faults in a distribution network disclosed in an embodiment of the present application; Figure 2 It is a module schematic diagram of a distribution network switch fault intelligent diagnosis device disclosed in an embodiment of the present application; Figure 3 It is a structural schematic diagram of an electronic device disclosed in an embodiment of the present application.

[0027] Explanation of the reference numerals: 201, acquisition module; 202, processing module; 300, electronic device; 301, processor; 302, communication bus; 303, user interface; 304, network interface; 305, memory. DETAILED DESCRIPTION

[0028] In order to enable technicians in this field to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments.

[0029] In the description of the embodiments of the present application, words such as "for example" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "for example" or "for example" is intended to present related concepts in a specific way.

[0030] In the description of the embodiments of the present application, the meaning of the term "multiple" refers to two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "include", "comprise", "have" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

[0031] The present application provides a method for intelligent diagnosis of switch faults in a distribution network, referring to Figure 1 , Figure 1 This is a flow chart of a method for intelligent diagnosis of distribution network switch faults provided by an embodiment of the present application. The method is applied to a server, which is a server that executes a distribution network switch fault intelligent diagnosis program. The server can be a single server, or a server cluster composed of multiple servers, or a cloud computing service center. The method includes steps S101 to S107, and the above steps are as follows: Step S101: Acquire operating status data of a distribution network switch, where the operating status data includes current data, voltage data, temperature data, and vibration data of the distribution network switch.

[0032] In step S101, various sensors, such as current sensors, voltage sensors, temperature sensors, and vibration sensors, are installed on the distribution network switch site. These sensors collect the operating status data of the distribution network switch in real time, and transmit the data to the server through a wired or wireless communication network. The server receives the operating status data, including current data, voltage data, temperature data, and vibration data.

[0033] Step S102: construct a multi-dimensional feature template of the distribution network switch in a normal state, the multi-dimensional feature template includes multiple feature templates, the feature template includes a current feature template, a voltage feature template, a temperature feature template and a vibration feature template, and the multi-dimensional feature template is trained by historical operating status data.

[0034] In step S102, a multi-dimensional feature template of the distribution network switch in a normal state is constructed, specifically including: acquiring historical operating status data of multiple normally operating distribution network switches to obtain a historical data set, the historical data set including historical current data, historical voltage data, historical temperature data and historical vibration data; performing a preprocessing operation on the historical data set to obtain a target historical data set; the preprocessing operation includes outlier removal, data normalization and data smoothing; using a sliding time window method, on different time scales, extracting multi-domain features of the target historical data set to obtain a feature sample set corresponding to each historical operating status data, the multi-domain features including time domain features, frequency domain features and time-frequency domain features; based on the feature sample sets corresponding to each historical operating status data, respectively training support vector data description models to obtain current feature templates, voltage feature templates, temperature feature templates and vibration feature templates.

[0035] Specifically, the server obtains the historical operating status data of multiple normally operating distribution network switches to form a historical data set. These data can come from the historical archives of the distribution network operation and maintenance department, or can be accumulated over a long period of time through a real-time data acquisition system. The acquired historical data sets often have quality problems, such as missing data, outliers, and noise. Therefore, the server performs preprocessing operations on the historical data sets, and the preprocessing operations include: outlier removal, data normalization, and data smoothing; for outlier removal, the server uses statistical methods (such as the 3σ principle) or machine learning algorithms (such as isolation forests) to detect and remove outliers in the data; for data normalization, the server uniformly scales data of different dimensions to the same scale (such as the 0-1 interval) to eliminate the impact of dimension differences on feature extraction. For data smoothing, the server uses filtering algorithms (such as moving average, Kalman filtering) to smooth the data and remove high-frequency noise. The preprocessed historical data set is called the target historical data set, which is used for subsequent feature extraction and template construction.

[0036] The server extracts features from the target historical data set obtained after preprocessing. Feature extraction is a key step in building feature templates. The server uses a sliding time window method to extract multi-domain features of the target historical data set at different time scales. Specifically: For time domain features, the server extracts statistical features of the data, such as mean, variance, peak-to-peak value, and root mean square, to reflect the overall distribution of the data. For frequency domain features, the server performs Fourier transform on the data to extract spectral features of the data, such as frequency, amplitude, and phase, to reflect the periodicity and frequency distribution of the data. For time-frequency domain features, the server performs wavelet transform or Hilbert-Huang transform on the data to extract the joint time-frequency features of the data, such as wavelet coefficients and instantaneous frequency, to reflect the non-stationary characteristics of the data. Through feature extraction, the server obtains the feature sample set corresponding to each historical operating status data (current, voltage, temperature, vibration).

[0037] Based on the feature sample set, the server trains the support vector data description (SVDD) model to obtain the corresponding feature template; for each feature sample set, the server trains an SVDD model to learn the optimal hypersphere of normal data so that it can tightly surround most normal data points. Finally, the server can obtain feature templates in four feature dimensions: current, voltage, temperature, and vibration, which together constitute the multi-dimensional feature template of the distribution network switch, including current feature template, voltage feature template, temperature feature template, and vibration feature template.

[0038] For example, suppose that the server collects operating status data from 10 normally operating distribution network switches for one month, with a data sampling frequency of once per minute. In the preprocessing stage, the server removes individual missing or abnormal data points and normalizes the maximum and minimum values ​​of the data in each dimension. Then, the server sets the time window size to 1 hour, extracts 20 time domain features, 10 frequency domain features, and 5 time-frequency domain features in each time window, and finally obtains feature samples of each switch at different time points. Next, the server uses these feature samples to train the SVDD model of current, voltage, temperature, and vibration features, and optimizes the model hyperparameters through cross-validation and grid search to obtain a feature template that describes the normal operating status. The feature template can provide a reference for subsequent distribution network switch fault diagnosis.

[0039] Step S103: Calculate the similarity between the running status data and each feature template to obtain multiple similarity values.

[0040] In step S103, the similarity between the operating status data and each feature template is calculated to obtain multiple similarity values, specifically including: determining the corresponding features of the operating status data relative to the feature template to obtain the corresponding feature sequence to be measured, the feature sequence to be measured includes a current feature sequence, a voltage feature sequence, a temperature feature sequence and a vibration feature sequence; calculating the optimal matching distance between the feature sequence to be measured and the feature template; within a preset time window, calculating the average value of the optimal matching distance to obtain the average matching distance, and performing an exponential transformation on the average matching distance to obtain a similarity value.

[0041] Specifically, the server extracts the feature sequence to be tested corresponding to the feature template from the real-time collected operating status data of the distribution network switch. Specifically, for the four physical quantities of current, voltage, temperature, and vibration, their time domain, frequency domain, and time-frequency domain features are extracted respectively to form four feature sequences to be tested. This step is the same as the feature extraction method when constructing the feature template, which ensures the comparability between the data to be tested and the feature template.

[0042] For each feature sequence to be tested, the server calculates the optimal matching distance between it and the corresponding feature template. The optimal matching distance can be used to find the optimal nonlinear alignment between two time series and measure their shape similarity. After obtaining the optimal matching distance, the server calculates the average value of the optimal matching distance within a preset time window (such as 1 minute) as the average matching distance between the switch operating state and the feature template within the time window. The smaller the average matching distance, the closer the switch operating state is to the normal state. In order to map the average matching distance to a bounded similarity value, the server performs an exponential transformation on the average matching distance to obtain the corresponding similarity value.

[0043] In a possible implementation, the calculation formula for calculating the optimal matching distance between the feature sequence to be tested and the feature template is: ;

[0044] Where D(i, j) represents the cumulative distance between the first i elements of the feature sequence to be tested and the first j elements of the feature template, x i Represents the i-th element of the feature sequence to be tested, y j represents the jth element of the feature template, (x i -y j ) 2 It represents the Euclidean distance between the i-th element of the feature sequence to be tested and the j-th element of the feature template.

[0045] Specifically, the above formula finds the optimal matching path between two sequences (the feature sequence to be tested and the feature template) by stretching and shifting on the time axis, so that the cumulative distance between the matching point pairs is minimized. Suppose the feature sequence to be tested is X=(x 1 , x 2 , ..., x n ), the feature template sequence is Y=(y 1 ,y 2 , ..., y m), define the distance matrix D, where D(i, j) represents the cumulative distance between the first i elements of X and the first j elements of Y, that is, the optimal matching distance. According to the idea of ​​dynamic programming, when calculating D(i, j), we can use the optimal solutions of the subproblems that have been calculated, namely D(i-1, j), D(i-1, j-1) and D(i, j-1), plus the current matching point pair (x i ,y j ), thus obtaining the recursive formula of D(i, j): D(i, j)=min{D(i-1, j), D(i-1, j-1), D(i, j-1)}+(x i -y j ) 2 ; Among them, (x i -y j ) 2 Represents x i and j The square of the Euclidean distance between them, that is, the local distance measure between the matching point pairs. The meaning of the above recursive formula is: when calculating D(i, j), the optimal matching path can come from three directions: (i-1, j): represents x i With y j No match, the optimal path comes from D(i-1, j), that is, x i was skipped.

[0046] (i-1, j-1): represents x i With y j Matching, the optimal path comes from D(i-1, j-1), that is, x i and j is matched.

[0047] (i, j-1): represents x i With y j No match, the optimal path comes from D(i, j-1), that is, y j was skipped.

[0048] The server selects the one with the smallest cumulative distance among the three directions, and adds the distance of the current matching point pair to obtain the optimal solution of D(i, j). The boundary conditions are: D(0, 0) = 0, D(i, 0) = D(0, j) = ∞ (infinity), that is, the distance between the empty sequence and any sequence is 0, and the distance between any sequence and the empty sequence is ∞. The distance matrix D is filled by dynamic programming, and finally D(n, m) is the optimal matching distance between the feature sequence X to be tested and the feature template sequence Y, that is, the similarity measure between the two sequences.

[0049] For example, assuming that the feature sequence to be tested is X=(1, 2, 3, 4, 5) and the feature template sequence is Y=(2, 3, 4), the calculation process of the distance matrix D is as follows: Initialize D(0,0)=0; D(i,0)=D(0,j)=∞, i=1,2,...,5; j=1,2,3; calculate the first row D(1,1)=min{∞,∞,∞}+(1-2) 2 =1, D(1,2)=min{∞,1,∞}+(1-3) 2 =5, D(1,3)=min{∞,5,∞}+(1-4) 2 =14; calculate the second row D(2,1)=min{1,∞,∞}+(2-2) 2 =1, D(2,2)=min{5,1,1}+(2-3) 2 =2, D(2,3)=min{14,2,5}+(2-4) 2 =6; And so on, until the entire distance matrix is ​​filled: ; Finally, D(5,3)=6 is the optimal matching distance between the feature sequence to be tested X and the feature template sequence Y. By backtracking the distance matrix, the optimal matching path can be obtained as ((1,2), (2,3), (4,3), (5,3)), and the corresponding matching relationships are x1y2, x2y3, x4y3, x5y3.

[0050] In a possible implementation, within a preset time window, the average value of the optimal matching distance is calculated, and a specific calculation formula for obtaining the average matching distance is:

[0051] in, is the average matching distance, n is the number of preset time windows, is the optimal matching distance in the kth time window.

[0052] Specifically, in the above formula, the server calculates the average value of the optimal matching distance within the preset time window to obtain the average matching distance. represents the average matching distance, n represents the number of preset time windows, The superscript (k) represents the optimal matching distance in the kth time window.

[0053] The meaning of this formula is: within a preset time window, the optimal matching distance is calculated for each time window, and then these optimal matching distances are added together and divided by the number of time windows to obtain the average matching distance.

[0054] For example, assuming that the preset time window is 5 minutes, within these 5 minutes, the optimal matching distance is calculated once every minute, and the obtained optimal matching distance sequence is {2.3, 1.9, 2.5, 2.1, 2.4}, then the calculation process of the average matching distance is as follows: =(2.3+1.9+2.5+2.1+2.4) / 5=2.24 In a possible implementation, the average matching distance is subjected to exponential transformation to obtain a specific calculation formula for the similarity value:

[0055] Among them, S is the similarity value, λ is the similarity adjustment factor, is the average matching distance.

[0056] Specifically, in the above formula, the server performs an exponential transformation on the average matching distance to obtain a similarity value. In the formula, S represents the similarity value, λ represents the similarity adjustment factor, Represents the average matching distance. The meaning of this formula is: first, take the negative value of the average matching distance, then multiply it by a similarity adjustment factor λ, and finally raise the product to the exponential of the natural constant e to get the similarity value.

[0057] The purpose of the exponential transformation is to map the average matching distance to the interval (0, 1] to obtain a normalized similarity measure. Since the smaller the average matching distance, the more similar the two sequences are, it is necessary to take a negative value for it, and then transform it into a similarity value that is closer to 1 and closer to 0, which is less similar, through the exponential transformation. The role of the similarity adjustment factor λ is to control the rate and shape of the exponential transformation. The larger the λ, the faster the rate of the exponential transformation, and the similarity value will quickly approach 1 when the distance is small; the smaller the λ, the slower the rate of the exponential transformation, and the similarity value will maintain a certain degree of distinction when the distance is large. By adjusting the size of λ, a balance can be achieved between matching accuracy and robustness.

[0058] For example, assuming the average matching distance =2.24, similarity adjustment factor λ=0.5, then the similarity value calculation process is as follows: S=e^(-0.5*2.24)=e^(-1.12)≈0.326 It can be seen that when the average matching distance is 2.24, the similarity value is 0.326, indicating that the similarity between the two sequences is relatively low. If λ=1, the similarity value will become e^(-2.24)≈0.106. It can be seen that the increase of λ will accelerate the decay rate of the similarity value. The value of λ can be designed according to the actual application scenario requirements, and this application does not limit this.

[0059] Step S104: multiple similarity values ​​are integrated to obtain a comprehensive health index of the distribution network switch.

[0060] In step S104, multiple similarity values ​​are integrated to obtain the specific formula of the comprehensive health index of the distribution network switch:

[0061] Among them, H is the comprehensive health index, N is the number of feature templates, and w i is the weight of the i-th feature template, S i is the similarity value of the i-th feature template, α i is the exponential coefficient for adjusting the i-th feature template.

[0062] Specifically, in the health status assessment of distribution network switches, multiple feature dimensions are usually considered comprehensively, such as current, voltage, temperature, and vibration, and each dimension has a corresponding feature template. By calculating the similarity between the real-time operation data of the switch and each feature template, a set of similarity values ​​{S 1 , S 2 , ..., S n}, where N is the number of feature templates. In order to merge these similarity values ​​into a comprehensive health index, the weighted exponential average method is adopted considering the importance differences of different feature dimensions and the nonlinear influence of similarity values. In the formula, H represents the comprehensive health index, w i represents the weight of the i-th feature template, S i Represents the similarity value of the i-th feature template, α i Represents the exponential coefficient for adjusting the i-th feature template. The numerator of the formula is to perform an exponential transformation on the similarity value of each feature template, then multiply it by the corresponding weight, and then sum it. Among them, the role of the exponential transformation is to map the similarity value Si from the linear space to the exponential space. By adjusting α i The size of α can control the nonlinearity of the mapping. i The larger the value, the stronger the nonlinearity of the exponential transformation, and the slight differences in similarity values ​​will be amplified; i The smaller it is, the weaker the nonlinearity of the exponential transformation is, and the difference in similarity values ​​will be compressed.

[0063] Weight w i The role of is to reflect the importance of the i-th feature template in the comprehensive evaluation. i The larger the value is, the greater the contribution of the i-th feature template to the comprehensive health index is; iThe smaller it is, the smaller the influence of the i-th feature template. By setting the weights reasonably, key features can be highlighted, minor features can be suppressed, and the pertinence of the comprehensive evaluation can be improved. Some features (such as current and voltage) may be more critical to the health status of the equipment, so they can be given higher weights, which is not limited in this application. The denominator of the formula is the sum of all weights, which plays a normalization role to ensure that the value range of the comprehensive health index H is between [0, 1].

[0064] Step S105: Compare the comprehensive health index with a preset fault threshold to determine whether a distribution network switch fails.

[0065] In step S105, the server compares the calculated comprehensive health index with the preset fault threshold to determine whether the switch is faulty. The preset fault threshold is an empirical value that can be set based on historical data and expert knowledge, indicating the dividing point between a healthy state and a faulty state. This application does not limit the specific value of the preset fault threshold.

[0066] Step S106: If it is determined that the comprehensive health index is greater than or equal to the preset fault threshold, it is determined that the distribution network switch has not failed.

[0067] In step S106, when the server determines that the comprehensive health index is greater than or equal to the preset fault threshold, it indicates that the operating status of the switch is generally normal and no significant fault signs occur.

[0068] Step S107: If it is determined that the comprehensive health index is less than the preset fault threshold, it is determined that the distribution network switch has not failed.

[0069] In step S107, when the server determines that the comprehensive health index is less than the preset fault threshold, it indicates that the operating state of the distribution network switch is significantly deviated from the normal operating condition and a fault has occurred. After determining the fault, the server generates a fault report and triggers the corresponding alarm and maintenance work order to notify the operation and maintenance personnel to handle it as soon as possible to avoid the expansion of the fault.

[0070] Reference Figure 2The present application also provides a distribution network switch fault intelligent diagnosis device, which is a server, and the server includes an acquisition module 201 and a processing module 202, wherein: the acquisition module 201 is used to obtain the operating status data of the distribution network switch, and the operating status data includes the current data, voltage data, temperature data and vibration data of the distribution network switch; the processing module 202 is used to construct a multi-dimensional feature template of the distribution network switch in a normal state, and the multi-dimensional feature template includes multiple feature templates, and the feature template includes a current feature template, a voltage feature template, a temperature feature template and a vibration feature template. The multi-dimensional feature template is composed of historical operating status The processing module 202 is also used to calculate the similarity between the operating status data and each feature template to obtain multiple similarity values; the processing module 202 is also used to fuse multiple similarity values ​​to obtain the comprehensive health index of the distribution network switch; the processing module 202 is also used to compare the comprehensive health index with the preset fault threshold to determine whether the distribution network switch has a fault; the processing module 202 is also used to determine that the distribution network switch has not failed if it is determined that the comprehensive health index is greater than or equal to the preset fault threshold; the processing module 202 is also used to determine that the distribution network switch has a fault if it is determined that the comprehensive health index is less than the preset fault threshold.

[0071] In a possible implementation, the processing module 202 constructs a multi-dimensional feature template of the distribution network switch in a normal state, specifically including: the acquisition module 201 acquires the historical operating status data of multiple normally operating distribution network switches to obtain a historical data set, and the historical data set includes historical current data, historical voltage data, historical temperature data and historical vibration data; the processing module 202 performs a preprocessing operation on the historical data set to obtain a target historical data set; the preprocessing operation includes outlier removal, data normalization and data smoothing; the processing module 202 uses a sliding time window method to extract multi-domain features of the target historical data set at different time scales to obtain a feature sample set corresponding to each historical operating status data, and the multi-domain features include time domain features, frequency domain features and time-frequency domain features; the processing module 202 trains a support vector data description model based on the feature sample sets corresponding to each historical operating status data to obtain a current feature template, a voltage feature template, a temperature feature template and a vibration feature template.

[0072] In a possible implementation, the processing module 202 calculates the similarity between the operating status data and each feature template to obtain multiple similarity values, specifically including: the processing module 202 determines the corresponding features of the operating status data relative to the feature template to obtain the corresponding feature sequence to be measured, the feature sequence to be measured includes a current feature sequence, a voltage feature sequence, a temperature feature sequence and a vibration feature sequence; the processing module 202 calculates the optimal matching distance between the feature sequence to be measured and the feature template; the processing module 202 calculates the average value of the optimal matching distance within a preset time window to obtain the average matching distance, and performs an exponential transformation on the average matching distance to obtain a similarity value.

[0073] In a possible implementation, the processing module 202 calculates the optimal matching distance between the feature sequence to be tested and the feature template using the following calculation formula: ; Where D(i, j) represents the cumulative distance between the first i elements of the feature sequence to be tested and the first j elements of the feature template, x i Represents the i-th element of the feature sequence to be tested, y j represents the jth element of the feature template, (x i -y j ) 2 It represents the Euclidean distance between the i-th element of the feature sequence to be tested and the j-th element of the feature template.

[0074] In a possible implementation, the processing module 202 calculates the average value of the optimal matching distance within a preset time window, and obtains a specific calculation formula of the average matching distance as follows: ; in, is the average matching distance, n is the number of preset time windows, is the optimal matching distance in the kth time window.

[0075] In a possible implementation, the processing module 202 performs an exponential transformation on the average matching distance to obtain a specific calculation formula for the similarity value: ; Among them, S is the similarity value, λ is the similarity adjustment factor, is the average matching distance.

[0076] In a possible implementation manner, the processing module 202 fuses multiple similarity values ​​to obtain a specific formula for the comprehensive health index of the distribution network switch: ; Among them, H is the comprehensive health index, N is the number of feature templates, and wi is the weight of the i-th feature template, S i is the similarity value of the i-th feature template, α i is the exponential coefficient for adjusting the i-th feature template.

[0077] It should be noted that: when the device provided in the above embodiment realizes its function, only the division of the above functional modules is used as an example. In actual application, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0078] The present application also provides an electronic device. Figure 3 , Figure 3 The electronic device 300 may include: at least one processor 301 , at least one network interface 304 , a user interface 303 , a memory 305 , and at least one communication bus 302 .

[0079] The communication bus 302 is used to realize the connection and communication between these components.

[0080] The user interface 303 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 303 may also include a standard wired interface and a wireless interface.

[0081] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0082] Among them, the processor 301 may include one or more processing cores. The processor 301 uses various interfaces and lines to connect various parts in the entire server, and executes various functions of the server and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 305, and calling data stored in the memory 305. Optionally, the processor 301 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). The processor 301 can integrate one or a combination of a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 301, and it can be implemented separately through a chip.

[0083] Among them, the memory 305 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read-Only Memory). Optionally, the memory 305 includes a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 305 may optionally also be at least one storage device located away from the aforementioned processor 301. Refer to Figure 3 , the memory 305 as a computer storage medium may include an operating system, a network communication module, a user interface module and an application program of an intelligent diagnosis method for distribution network switch faults.

[0084] exist Figure 3In the electronic device 300 shown, the user interface 303 is mainly used to provide an input interface for the user and obtain the data input by the user; and the processor 301 can be used to call the application program stored in the memory 305 for a method for intelligent diagnosis of distribution network switch faults. When executed by one or more processors 301, the electronic device 300 executes one or more of the methods described in the above embodiments. It should be noted that for the aforementioned method embodiments, for the sake of simple description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited to the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required for the present application.

[0085] The present application also provides a computer-readable storage medium, which stores instructions. When executed by one or more processors 301, the electronic device 300 executes one or more of the methods described in the above embodiments.

[0086] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0087] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0088] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0089] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0090] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a memory and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned memory includes: various media that can store program codes, such as USB flash drives, mobile hard drives, magnetic disks or optical disks.

[0091] The above is only an exemplary embodiment of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification and the truth of practice, those skilled in the art will easily think of other embodiments of the present disclosure.

[0092] This application is intended to cover any variation, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the art not described in the present disclosure. The description and examples are to be regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A distribution network switch fault intelligent diagnosis method, characterized in that: The method comprises: Acquire operating status data of the distribution network switch, wherein the operating status data includes current data, voltage data, temperature data and vibration data of the distribution network switch; Constructing a multi-dimensional feature template of the distribution network switch in a normal state, wherein the multi-dimensional feature template includes a plurality of feature templates, including a current feature template, a voltage feature template, a temperature feature template, and a vibration feature template, and the multi-dimensional feature template is obtained by training historical operating status data; Calculating the similarity between the running status data and each of the feature templates to obtain a plurality of similarity values; Fusion of the plurality of similarity values ​​to obtain a comprehensive health index of the distribution network switch; Comparing the comprehensive health index with a preset fault threshold to determine whether the distribution network switch has a fault; If it is determined that the comprehensive health index is greater than or equal to the preset fault threshold, it is determined that the distribution network switch has not failed; If it is determined that the comprehensive health index is less than the preset fault threshold, it is determined that the distribution network switch has a fault.

2. The method according to claim 1, characterized in that The constructing of a multi-dimensional feature template of the distribution network switch in a normal state specifically includes: Acquire historical operating status data of a plurality of normally operating distribution network switches to obtain a historical data set, wherein the historical data set includes historical current data, historical voltage data, historical temperature data, and historical vibration data; Performing a preprocessing operation on the historical data set to obtain a target historical data set; the preprocessing operation includes outlier removal, data normalization, and data smoothing; By using a sliding time window method, multi-domain features of the target historical data set are extracted at different time scales to obtain feature sample sets corresponding to each of the historical operating status data, wherein the multi-domain features include time domain features, frequency domain features, and time-frequency domain features; Based on the feature sample sets corresponding to the historical operating status data, support vector data description models are trained respectively to obtain the current feature template, the voltage feature template, the temperature feature template and the vibration feature template.

3. The method according to claim 1, characterized in that The calculating of the similarity between the running status data and each feature template to obtain a plurality of similarity values ​​specifically includes: Determine the corresponding features of the operating state data relative to the feature template to obtain a corresponding feature sequence to be measured, wherein the feature sequence to be measured includes a current feature sequence, a voltage feature sequence, a temperature feature sequence, and a vibration feature sequence; Calculating the optimal matching distance between the feature sequence to be tested and the feature template; In a preset time window, the average value of the optimal matching distance is calculated to obtain an average matching distance, and the average matching distance is exponentially transformed to obtain a similarity value.

4. The method according to claim 3, characterized in that The calculation formula for calculating the optimal matching distance between the feature sequence to be tested and the feature template is: ； Where D(i, j) represents the cumulative distance between the first i elements of the feature sequence to be tested and the first j elements of the feature template, x i represents the i-th element of the feature sequence to be tested, y j represents the jth element of the feature template, (x i -y j ) 2 Represents the Euclidean distance between the i-th element of the feature sequence to be tested and the j-th element of the feature template.

5. The method according to claim 3, characterized in that: The specific calculation formula for calculating the average value of the optimal matching distance within the preset time window is: ； in, is the average matching distance, n is the number of the preset time windows, is the optimal matching distance in the kth time window.

6. The method according to claim 3, characterized in that The specific calculation formula for obtaining the similarity value by performing exponential transformation on the average matching distance is: ； Wherein, S is the similarity value, λ is the similarity adjustment factor, is the average matching distance.

7. The method according to claim 1, characterized in that The specific formula for fusing the multiple similarity values ​​to obtain the comprehensive health index of the distribution network switch is: ； Wherein, H is the comprehensive health index, N is the number of feature templates, and w i is the weight of the i-th feature template, S i is the similarity value of the i-th feature template, α i is the exponential coefficient for adjusting the i-th feature template.

8. An intelligent diagnostic device for switch faults in a distribution network, characterized in that: The device comprises an acquisition module (201) and a processing module (202), wherein: The acquisition module (201) is used to acquire operating status data of the distribution network switch, wherein the operating status data includes current data, voltage data, temperature data and vibration data of the distribution network switch; The processing module (202) is used to construct a multi-dimensional feature template of the distribution network switch in a normal state, the multi-dimensional feature template comprising a plurality of feature templates, the feature template comprising a current feature template, a voltage feature template, a temperature feature template and a vibration feature template, the multi-dimensional feature template being obtained by training historical operating state data; The processing module (202) is further used to calculate the similarity between the running status data and each of the feature templates to obtain a plurality of similarity values; The processing module (202) is further used to fuse the multiple similarity values ​​to obtain a comprehensive health index of the distribution network switch; The processing module (202) is further used to compare the comprehensive health index with a preset fault threshold to determine whether the distribution network switch has a fault; The processing module (202) is further configured to determine that the distribution network switch has not failed if it is determined that the comprehensive health index is greater than or equal to the preset fault threshold; The processing module (202) is further configured to determine that a fault has occurred in the distribution network switch if it is determined that the comprehensive health index is less than the preset fault threshold.

9. An electronic device, characterized in that: The electronic device (300) comprises a processor (301), a memory (305), a user interface (303) and a network interface (304), wherein the memory (305) is used to store instructions, the user interface (303) and the network interface (304) are used to communicate with other devices, and the processor (301) is used to execute the instructions stored in the memory (305) so that the electronic device (300) executes the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed, the method according to any one of claims 1 to 7 is performed.

Images