Fault studying and judging method and system based on abnormal signals

Abstract

The invention discloses a fault studying and judging method and system based on an abnormal signal, and relates to the technical field of main network abnormal signal data analysis, and the method comprises the steps: collecting main network abnormal signal data, forming a data set, and carrying out the initialization of a clustering center through a k-means clustering K-Means + + algorithm; calculating the distance between a data point in the data set and the clustering center, and distributing the data point to a cluster corresponding to the nearest clustering center; updating a clustering center according to the cluster and carrying out iteration; and evaluating the performance of the k-mean clustering K-Means + + algorithm by using the contour coefficient. The method comprises the following steps: extracting main network abnormal signal data and performing machine learning by using a k-means clustering K-Means + + algorithm according to a sequence of sending abnormal signals when a power grid has a fault and a correlation rule between the abnormal signals and associated signals after the fault occurs, and optimizing an output result, namely a researched and judged fault result, by operating the algorithm for multiple times. Therefore, fault study and judgment based on the abnormal signals are carried out, and the study and judgment accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of main network abnormal signal data analysis, and in particular to a fault analysis method and system based on abnormal signals. Background Art

[0002] With the rapid development of Yunnan power grid dispatching business, the number of power plants and substations continues to increase, making the information and work tasks that controllers need to monitor and process become more and more onerous; with the large-scale growth of abnormal signal data in the power grid, it has become an inevitable trend to combine big data analysis technology with power grid control business to realize the development of intelligent main grid control business. This trend is not only to meet the needs of expanding business scale, but also to improve the efficiency and accuracy of controllers in processing information; however, despite the broad application prospects of big data analysis technology, the existing main grid control business monitoring system has not yet kept up with this development pace. The system is still in the statistical analysis stage and fails to combine algorithms with big data to realize intelligent main grid control, which has led to the potential value of the existing main grid abnormal signal data not being fully utilized; in order to overcome this limitation, system upgrades and improvements are inevitable, and the introduction of more advanced technical means to enable it to more accurately analyze and judge abnormal power grid signal data is a problem that needs to be solved urgently.

[0003] The common solutions currently available have many shortcomings, including: the existing control data analysis systems mostly use the SpringCloud framework, Echarts technology, and Bootstrap technology, but the analysis of data remains at the simple statistical analysis level of surface graphics and reports, and fails to deeply mine and analyze the potential information of the data, resulting in the controllers failing to obtain the information they urgently need, and the analysis and utilization of abnormal signal data are insufficient; in terms of technology, the existing systems use some algorithms for fault analysis, but these algorithms do not combine the abnormal signal data within the time period, resulting in deviations in the analysis results. Summary of the invention

[0004] In view of the fact that when monitoring and processing power grid dispatching business, the existing technology does not analyze and utilize abnormal signal data to a sufficient extent, and the abnormal signal data within the time period is not combined during the judgment, resulting in deviations in the judgment results, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is how to provide a method for improving the accuracy of fault analysis by combining abnormal signal data within a time period.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In the first aspect, an embodiment of the present invention provides a fault analysis method based on abnormal signals, which includes collecting abnormal signal data of the main network and forming a data set, initializing the cluster center using the k-means clustering K-Means++ algorithm; calculating the distance between the data points in the data set and the cluster center, and assigning the data points to the cluster corresponding to the cluster center closest to the data point; updating the cluster center according to the cluster and iterating; and evaluating the performance of the k-means clustering K-Means++ algorithm using the silhouette coefficient.

[0008] As a preferred solution of the fault analysis method based on abnormal signals described in the present invention, wherein: the main network abnormal signal data includes abnormal signals at the time of fault, abnormal signals before fault and abnormal signals after fault; the initialization of cluster centers using the k-means clustering K-Means++ algorithm includes the following steps: randomly selecting a data point in the data set as the first cluster center; calculating the probability of a data point that is not selected as a cluster center being selected as the next cluster center, and selecting the next cluster center; repeating the above steps of calculating the probability of a data point that is not selected as a cluster center being selected as the next cluster center until k cluster centers are selected.

[0009] As a preferred solution of the fault analysis method based on abnormal signals of the present invention, the calculation of the probability of a data point that is not selected as a cluster center being selected as the next cluster center includes the following steps: calculating the shortest distance between the unselected data point and the selected cluster center; calculating the probability of the unselected data point being selected as the next cluster center according to the shortest distance; the specific formula for the probability of the unselected data point being selected as the next cluster center is as follows:

[0010]

[0011] Among them, P(x) is the probability of a data point being selected as the next cluster center; x is a data point that has not been selected; is the square of the distance from the unselected data point x to the nearest cluster center; σ is the parameter that controls the shape of the probability distribution; N is the total number of data points in the data set; is the square of the distance from the unselected data point j to the nearest cluster center.

[0012] As a preferred solution of the fault analysis method based on abnormal signals of the present invention, the specific formula for calculating the distance between the data points in the data set and the cluster center is as follows:

[0013]

[0014] Among them, x i is all the data points in the data set; c jFor all cluster centers; distance(x i ,c j ) is the distance between the data point in the data set and the cluster center; n is the number of cluster centers; x i,k is the kth data point; c j,k is the kth cluster center.

[0015] As a preferred solution of the abnormal signal-based fault analysis method of the present invention, the specific formula for allocating data points to the cluster corresponding to the nearest cluster center is as follows:

[0016] j=argminjdistance(,cj)

[0017] Among them, j is the index of the cluster center, indicating the cluster center c to which the data point xi is assigned j The index of the corresponding cluster; c j is the center of the jth cluster.

[0018] As a preferred solution of the abnormal signal-based fault analysis method of the present invention, the cluster center is updated by taking the mean of all data points in the cluster as the new cluster center; the specific formula for updating the cluster center is as follows:

[0019]

[0020] Among them, C j is the cluster center c j The corresponding cluster; the stopping condition of the iteration is the maximum number of iterations, the cluster center no longer changes or the square error within the cluster.

[0021] As a preferred solution of the fault analysis method based on abnormal signals of the present invention, the contour coefficient includes cohesion and separation; the cohesion is the closeness between a sample point and the elements within the class; the separation is the closeness between a sample point and the elements outside the class; the specific formula of the contour coefficient is as follows:

[0022]

[0023] Among them, S is the silhouette coefficient; a i is the cohesion of the ith data point, which represents the average distance from the corresponding data point to other points in the same cluster; b i is the separation degree of the i-th data point, which represents the average distance from the corresponding data point to all points in the nearest neighbor cluster; N is the total number of data points in the data set.

[0024] On the second aspect, in order to further solve the safety problems existing in fault analysis, the embodiment of the present invention provides a fault analysis system based on abnormal signals, which includes: a data processing module, which is used to collect the abnormal signal data of the main network and form a data set, and select the initial cluster center by using the k-means clustering K-Means++ algorithm; a data update module, which is used to calculate the distance between the data point and the cluster center, and assign the data point to the cluster corresponding to the nearest cluster center, and update the initial cluster center according to the cluster and iterate; an evaluation module, which is used to evaluate the performance of the k-means clustering K-Means++ algorithm by using the silhouette coefficient.

[0025] In a third aspect, an embodiment of the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of a fault analysis method based on abnormal signals as described in the first aspect of the present invention is implemented.

[0026] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of a fault analysis method based on abnormal signals as described in the first aspect of the present invention.

[0027] The beneficial effects of the present invention are as follows: according to the characteristics that abnormal signals when a power grid fault occurs are emitted in sequence and that the abnormal signals after the fault occur have a correlation with the accompanying signals, the present invention uses the k-means clustering K-Means++ algorithm to extract the abnormal signals when the fault occurs, the abnormal signals before the fault, and the abnormal signals after the fault in the historical data, and substitutes them into the k-means clustering K-Means++ algorithm for machine learning, and optimizes the output results by running the algorithm multiple times, that is, the judged fault results, so as to perform fault judgment based on abnormal signals and improve the accuracy of judgment. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0029] Figure 1 This is a flow chart for implementing the present invention in Example 1.

[0030] Figure 2 This is a diagram of the training process of the k-means clustering K-Means++ algorithm in Example 1.

[0031] Figure 3This is a comparison chart of the performance of the K-Means++ algorithm evaluated using the silhouette coefficient in Example 2.

[0032] Figure 4 This is a statistical chart of the correct situation in Example 2.

[0033] Figure 5 This is a statistical chart of the accuracy of the analysis in Example 2. DETAILED DESCRIPTION

[0034] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0035] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0036] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0037] Example 1

[0038] Reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides a fault analysis method based on abnormal signals, comprising the following steps:

[0039] S1: Collect the abnormal signal data of the main network and form a data set, and use the K-Means++ algorithm to initialize the clustering center.

[0040] Preferably, Figure 1The implementation process of the present invention is shown as follows. First, the abnormal signal data of the main network is collected and a data set is formed. The k-means clustering K-Means++ algorithm is used to determine the number of classifications K and randomly select a data point as the first cluster centroid. Then, the distance between each data point in the data set and the corresponding nearest cluster centroid is calculated, and then a new data point is selected as the new cluster centroid. The greater the distance, the greater the probability that the data point is selected as the cluster centroid. After the new cluster centroid is selected, it is determined whether the number of selected cluster centroids is K. If the number of cluster centroids is K, the standard k-means clustering K-Means++ algorithm is executed; if it is not greater than K, the step of calculating the distance between each data point in the data set and the corresponding nearest cluster centroid is re-executed.

[0041] Preferably, the main network abnormal signal data includes the abnormal signal at the time of the fault, the abnormal signal 15 minutes before the fault, and the abnormal signal 15 minutes after the fault.

[0042] Preferably, initializing the cluster center using the K-Means++ algorithm includes the following steps: randomly selecting a data point in the data set as the first cluster center.

[0043] Calculate the probability of a data point that has not been selected as a cluster center being selected as the next cluster center, and select the next cluster center.

[0044] Repeat the above steps of calculating the probability of a data point that has not been selected as a cluster center being selected as the next cluster center until k cluster centers are selected.

[0045] Furthermore, calculating the probability that a data point that has not been selected as a cluster center is selected as the next cluster center includes the following steps: calculating the shortest distance between the unselected data point and the selected cluster center using the Euclidean distance.

[0046] The probability of an unselected data point being selected as the next cluster center is calculated based on the shortest distance, where the probability of an unselected data point being selected as the next cluster center is proportional to the square of the distance from the corresponding data point to the selected cluster center.

[0047] Specifically, the specific formula for the probability that an unselected data point is selected as the next cluster center is as follows:

[0048]

[0049] Among them, P(x) is the probability of a data point being selected as the next cluster center; x is a data point that has not been selected; is the square of the distance from the unselected data point x to the nearest cluster center; σ is the parameter that controls the shape of the probability distribution. When σ is small, the probability distribution will be more concentrated and more sensitive to distance. When σ is large, the probability distribution will be smoother and the impact on distance will be weakened; N is the total number of data points in the data set; is the square of the distance from the unselected data point j to the nearest cluster center.

[0050] Furthermore, by using the k-means clustering K-Means++ algorithm to intelligently initialize the cluster centers, selecting the appropriate number of cluster centers and data points based on abnormal signals, and using a probability calculation method based on Euclidean distance to select the next cluster center, the stability of the algorithm is improved compared to the traditional K-means, avoiding falling into local optimality, effectively improving the efficiency and global convergence speed of the algorithm, being able to adapt to different situations, and improving the accurate calculation of the probability of unselected data points.

[0051] S2: Calculate the distance between the data points in the data set and the cluster center, and assign the data points to the cluster corresponding to the cluster center closest to them.

[0052] Preferably, the Euclidean distance is used to calculate the distance between the data points in the data set and the cluster center.

[0053] Specifically, the specific formula for calculating the distance between the data points in the data set and the cluster center is as follows:

[0054]

[0055] Among them, xi is all the data points in the data set; c j For all cluster centers; distance(xi,c j ) is the distance between the data point in the data set and the cluster center; n is the number of cluster centers; x i , k is the kth data point; c j , k is the kth cluster center.

[0056] Furthermore, the specific formula for assigning data points to the cluster corresponding to the nearest cluster center is as follows:

[0057] j=arg min,distance(x:,cj)

[0058] Among them, j is the index of the cluster center, indicating the cluster center c to which the data point xi is assigned j The index of the corresponding cluster; c j is the center of the jth cluster.

[0059] S3: Update the cluster centers according to the clusters and iterate.

[0060] Preferably, the cluster center is updated by taking the mean of all data points in the cluster as the new cluster center.

[0061] Specifically, the specific formula for updating the cluster center is as follows:

[0062]

[0063] Among them, C j is the cluster center c j The corresponding cluster.

[0064] Preferably, the stopping condition of the iteration is the maximum number of iterations, the cluster center no longer changes, or the intra-cluster square error.

[0065] Specifically, Figure 2 The figure shows the training process of the K-Means++ algorithm, which iterates by repeatedly calculating the distance between the data points in the data set and the cluster center, assigning the data points to the cluster corresponding to the nearest cluster center, and updating the cluster according to the cluster until the stopping condition is met.

[0066] Furthermore, after the iteration, the output result of the K-Means++ algorithm is the fault analysis result obtained by the present invention based on the abnormal signal.

[0067] S4: Use the silhouette coefficient to evaluate the performance of the K-Means++ algorithm.

[0068] Preferably, the silhouette coefficient includes cohesion and separation.

[0069] Specifically, cohesion is the closeness between a sample point and the elements within the class.

[0070] Specifically, the separation degree is the closeness between a sample point and out-of-class elements.

[0071] Specifically, the specific formula for evaluating the performance of the K-Means++ algorithm using the silhouette coefficient is as follows:

[0072]

[0073] Among them, S is the silhouette coefficient; a i is the cohesion of the ith data point, which indicates the average distance from the corresponding data point to other points in the same cluster. The smaller the distance, the tighter the cluster is. i is the separation degree of the i-th data point, which indicates the average distance from the corresponding data point to all points in the nearest neighbor cluster. The larger the value, the farther the distance from the adjacent cluster. N is the total number of data points in the data set.

[0074] Furthermore, the performance of the k-means clustering K-Means++ algorithm in fault analysis based on abnormal information is evaluated by using the silhouette coefficient, and the cohesion and separation are comprehensively considered to provide an objective performance measurement for the clustering results, so that users can have a more comprehensive understanding of the advantages and disadvantages of the clustering results, indicating the advantages of the present invention in fault analysis.

[0075] This embodiment also provides a fault analysis system based on abnormal signals, including: a data processing module, which is used to collect the abnormal signal data of the main network and form a data set, and use the k-means clustering K-Means++ algorithm to select the initial cluster center; a data update module, which is used to calculate the distance between the data point and the cluster center, and assign the data point to the cluster corresponding to the nearest cluster center, and update the initial cluster center according to the cluster and iterate; an evaluation module, which is used to evaluate the performance of the k-means clustering K-Means++ algorithm using the silhouette coefficient.

[0076] This embodiment also provides a computer device, which is suitable for a fault analysis method based on abnormal signals, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement a fault analysis method based on abnormal signals as proposed in the above embodiment.

[0077] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0078] The present embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, a fault analysis method based on abnormal signals as proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0079] In summary, according to the characteristics that abnormal signals when power grid faults occur are sent in sequence and the abnormal signals after the faults occur have a correlation with the accompanying signals, the present invention uses the k-means clustering K-Means++ algorithm to extract the abnormal signals at the time of the fault, the abnormal signals before the fault and the abnormal signals after the fault in the historical data, and substitutes them into the k-means clustering K-Means++ algorithm for machine learning. By running the algorithm multiple times to optimize the output results, that is, the judged fault results, fault judgment based on abnormal signals is performed, thereby improving the accuracy of judgment.

[0080] Example 2

[0081] Reference Figure 3 to Figure 5 , which is the second embodiment of the present invention. This embodiment is different from the first embodiment in that, in order to verify its beneficial effects, operating data and related instructions of the present invention in an actual environment are provided.

[0082] This example collects abnormal signals at the time of the fault, abnormal signals 15 minutes before the fault, and abnormal signals 15 minutes after the fault, and uses the k-means clustering K-Means++ algorithm to select the cluster center. The final fault analysis result is obtained by updating the cluster center and iterating multiple times. At the same time, the performance of the k-means clustering K-Means++ algorithm is evaluated using the silhouette coefficient, which more comprehensively demonstrates the advantages of the present invention in fault analysis.

[0083] like Figure 3The figure shows the performance of the k-means clustering K-Means++ algorithm in fault diagnosis evaluated by the silhouette coefficient. It can be seen from the figure that as the number of selected cluster centers gradually increases, the value of the silhouette coefficient gradually decreases, and the performance of the k-means clustering K-Means++ algorithm gradually improves, indicating that the present invention effectively improves the accuracy of fault diagnosis.

[0084] like Figure 4 and Figure 5 The figure shows the correct situation statistics and accuracy statistics for fault analysis in this example. It can be seen from the figure that the present invention can effectively distinguish faults and correct related data when facing a large amount of power grid data, combine abnormal signal data within a time period, deeply mine and analyze the potential information of the data, improve the accuracy of the final analysis result, and provide strong support for the operation and maintenance of the power grid.

[0085] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A fault analysis method based on abnormal signals, characterized in that: include: Collect the abnormal signal data of the main network and form a data set, and use the K-Means++ algorithm to initialize the cluster center; Calculating the distance between the data points in the data set and the cluster center, and assigning the data points to the cluster corresponding to the cluster center closest to the data point; Update the cluster center according to the cluster and iterate; The silhouette coefficient was used to evaluate the performance of the k-means clustering K-Means++ algorithm.

2. The fault analysis method based on abnormal signals according to claim 1, characterized in that: The main network abnormal signal data includes abnormal signals at the time of the fault, abnormal signals before the fault and abnormal signals after the fault; The initialization of cluster centers using the K-Means++ algorithm includes the following steps: Randomly select a data point in the data set as the first cluster center; Calculate the probability of a data point that is not selected as a cluster center being selected as the next cluster center, and select the next cluster center; Repeat the above steps of calculating the probability of a data point that has not been selected as a cluster center being selected as the next cluster center until k cluster centers are selected.

3. The fault analysis method based on abnormal signals according to claim 2, characterized in that: The calculation of the probability of a data point not selected as a cluster center being selected as the next cluster center comprises the following steps: Calculate the shortest distance between the unselected data points and the selected cluster centers; Calculate the probability of an unselected data point being selected as the next cluster center according to the shortest distance; The specific formula for the probability of the unselected data point being selected as the next cluster center is as follows: Among them, P(x) is the probability of a data point being selected as the next cluster center; x is a data point that has not been selected; is the square of the distance from the unselected data point x to the nearest cluster center; σ is the parameter that controls the shape of the probability distribution; N is the total number of data points in the data set; is the square of the distance from the unselected data point j to the nearest cluster center.

4. The fault analysis method based on abnormal signals according to claim 3 is characterized in that: The specific formula for calculating the distance between the data points in the data set and the cluster center is as follows: Among them, xi is all the data points in the data set; c j for all cluster centers; distance(xi,c j ) is the distance between the data point in the data set and the cluster center; n is the number of cluster centers; x i , k is the kth data point; c j , k is the kth cluster center.

5. The fault analysis method based on abnormal signals according to claim 4, characterized in that: The specific formula for assigning data points to the cluster corresponding to the nearest cluster center is as follows: j = argminjdistance(x:,cj) Where j is the index of the cluster center; c j is the center of the jth cluster.

6. The fault analysis method based on abnormal signals according to claim 5, characterized in that: The cluster center is updated by taking the mean of all data points in the cluster as the new cluster center; The specific formula for updating the cluster center is as follows: Among them, C j is the cluster center c j The corresponding clusters; The stopping condition of the iteration is the maximum number of iterations, the cluster center no longer changes, or the intra-cluster square error.

7. The fault analysis method based on abnormal signals according to claim 6, characterized in that: The silhouette coefficient includes cohesion and separation; The cohesion is the closeness between a sample point and the elements in the class; The separation degree is the closeness between a sample point and out-of-class elements; The specific formula of the silhouette coefficient is as follows: Among them, S is the silhouette coefficient; a i is the cohesion of the ith data point, which represents the average distance from the corresponding data point to other points in the same cluster; b i is the separation degree of the i-th data point, which represents the average distance from the corresponding data point to all points in the nearest neighbor cluster; N is the total number of data points in the data set.

8. A fault analysis system based on abnormal signals, based on the fault analysis method based on abnormal signals according to any one of claims 1 to 7, characterized in that: include, The data processing module is used to collect the abnormal signal data of the main network and form a data set, and select the initial clustering center using the K-Means++ algorithm; The data update module is used to calculate the distance between the data point and the cluster center, assign the data point to the cluster corresponding to the nearest cluster center, update the initial cluster center according to the cluster and iterate; Evaluation module, used to evaluate the performance of K-Means++ algorithm using silhouette coefficient.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of a fault analysis method based on abnormal signals described in any one of claims 1 to 7 are implemented.

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

Images