Transformer anomaly detection method and device, storage medium and computer equipment

Through the intelligent optimization algorithm, the density clustering model is optimized, and the transformer monitoring data is clustered and analyzed, which solves the problem of transformer abnormal detection in the power system, realizes timely evaluation and early warning of the abnormal state of the transformer, and improves detection accuracy.

CN118940166AInactive Publication Date: 2024-11-12EAST CHINA BRANCH OF STATE GRID CORP
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202410892474.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-04
Publication Date
2024-11-12
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing transformer abnormality detection methods are difficult to timely and effectively analyze the huge data in the unified information platform of the power system, resulting in poor online assessment and early warning capabilities for the abnormality risk of transformer.

Method used

An intelligent optimization algorithm is used to optimize the domain radius and the minimum number of points contained in the domain in the density clustering model to obtain the density clustering optimization model, and cluster the transformer monitoring data set based on this model to analyze the abnormal state.

Benefits of technology

It improves the accuracy and timeliness of transformer abnormality detection, and enhances the online evaluation and early warning capabilities of transformer abnormality status.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118940166A_ABST
    Figure CN118940166A_ABST
Patent Text Reader

Abstract

The invention discloses a transformer anomaly detection method and device, a storage medium and computer equipment, relates to the technical field of transformer anomaly analysis, and mainly aims to solve the problem that an abnormal state of a transformer cannot be timely and effectively analyzed in integrated data of a unified information platform of an electric power system; the method comprises the steps that an intelligent optimization algorithm is adopted to carry out optimization processing on the domain radius in a density clustering model and the minimum number of points contained in the domain, and a density clustering optimization model is obtained; acquiring a transformer monitoring data set, and performing clustering processing on the transformer monitoring data set by adopting the density clustering optimization model to obtain a data clustering result; and analyzing the abnormal state of the transformer based on the data clustering result to obtain an abnormal analysis result of the transformer.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of transformer abnormality analysis, and in particular to a transformer abnormality detection method and device, a storage medium, and a computer device. Background Art

[0002] With the development of data platforms in the power sector, data on power equipment status monitoring, production management, operation scheduling, environmental meteorology, etc. will gradually be integrated and shared on a unified information platform, thereby promoting the development of power equipment status evaluation, detection and prediction technology towards comprehensive analysis based on panoramic status. However, there are many factors that affect the operating status of power equipment. The explosive growth of status monitoring data in the future, coupled with the huge amount of information data such as power grid operation and meteorological environment that are closely related to the status of the equipment, will make it difficult for existing transformer anomaly detection methods to analyze and process these huge amounts of data in a timely and effective manner, resulting in difficulty in timely detection of transformer anomalies, which reduces the online assessment and early warning capabilities of transformer abnormality risks. Summary of the invention

[0003] In view of this, the present invention provides a transformer abnormality detection method and device, a storage medium, and a computer device, the main purpose of which is to solve the problem that the abnormal state of the transformer cannot be analyzed in a timely and effective manner in the integrated data of the unified information platform of the power system, resulting in poor online assessment and early warning capabilities of the transformer abnormal state risk.

[0004] According to one aspect of the present invention, a transformer abnormality detection method is provided, comprising:

[0005] An intelligent optimization algorithm is used to optimize the domain radius and the minimum number of points in the domain in the density clustering model to obtain a density clustering optimization model.

[0006] Acquire a transformer monitoring data set, and use the density clustering optimization model to perform clustering processing on the transformer monitoring data set to obtain a data clustering result;

[0007] The abnormal state of the transformer is analyzed based on the data clustering result to obtain a transformer abnormality analysis result.

[0008] Furthermore, the intelligent optimization algorithm is one of a decentralized search algorithm, a particle swarm algorithm and a northern goshawk algorithm.

[0009] Furthermore, when the intelligent optimization algorithm is the Northern Goshawk algorithm, the intelligent optimization algorithm is used to optimize the domain radius and the minimum number of points contained in the domain in the density clustering model to obtain a density clustering optimization model, including:

[0010] Initializing the positions of all northern goshawks in the northern goshawk algorithm to obtain a population matrix of the northern goshawk; the positions of the northern goshawks represent a set of field radii and a minimum number of points contained in the field;

[0011] The hunting process of the northern goshawk is divided into a first stage optimization search and a second stage optimization search, and the position of the northern goshawk is iteratively updated based on the first stage optimization search, the second stage optimization search and a preset number of iterations to obtain a target position;

[0012] Based on the target position, the domain radius and the minimum number of points contained in the domain in the density clustering optimization model are determined to obtain the density clustering optimization model after parameter optimization.

[0013] Further, the iterative updating process of the position of the northern goshawk based on the first stage optimization search, the second stage optimization search and a preset number of iterations to obtain a target position includes:

[0014] The first stage optimization represents the recognition and attack of the northern goshawk on the prey, including updating the position of the northern goshawk based on the fitness function in the density clustering model to obtain a first updated position;

[0015] The second stage optimization represents the northern goshawk chasing the escaped prey, including updating the first updated position based on the goshawk's attack radius to obtain a second updated position;

[0016] The first stage optimization search and the second stage optimization search are repeatedly executed based on the preset number of iterations, and the first updated position and the second updated position are iteratively updated to obtain the target position.

[0017] Furthermore, the density clustering optimization model is used to cluster the transformer monitoring data set to obtain a data clustering result, including:

[0018] Traversing each data point in the transformer monitoring data set, and determining the number of data in the field corresponding to each data point based on the optimized field radius;

[0019] Determine the data points whose number of data in the field is greater than the minimum number of points included in the field after optimization as the core objects in the transformer monitoring data set;

[0020] The data points that are density-reachable with each of the core objects are determined recursively, and the density-reachable data points and the corresponding core objects are divided into the same cluster to obtain the data clustering result.

[0021] Further, the abnormal state of the transformer is analyzed based on the data clustering result to obtain the transformer abnormality analysis result, including:

[0022] Based on the data clustering result, data points in the transformer monitoring data set that do not belong to any cluster are marked to obtain marked abnormal points;

[0023] Sorting the marked abnormal points according to sampling time to obtain an abnormal data sequence;

[0024] If the abnormal data sequence contains n consecutive sampling points, it is determined that the transformer is in an abnormal state.

[0025] Furthermore, the intelligent optimization algorithm is used to optimize the domain radius and the minimum number of points in the domain in the density clustering model to obtain a density clustering optimization model, including:

[0026] Adopting a plurality of the intelligent optimization algorithms to respectively optimize the parameters in the density clustering model, and obtaining the domain radius and the minimum number of points contained in the domain corresponding to each of the intelligent optimization algorithms;

[0027] Calculate the average value of the domain radius and the average value of the minimum number of included points in the domain based on the domain radius and the minimum number of included points in the domain corresponding to each of the intelligent optimization algorithms;

[0028] The density clustering optimization model is obtained based on the average value of the domain radius and the average value of the minimum number of included points in the domain.

[0029] According to another aspect of the present invention, a transformer abnormality detection device is provided, comprising:

[0030] The model optimization module is used to optimize the domain radius and the minimum number of points in the domain in the density clustering model using an intelligent optimization algorithm to obtain a density clustering optimization model;

[0031] A clustering module, used to obtain a transformer monitoring data set, and perform clustering processing on the transformer monitoring data set using the density clustering optimization model to obtain a data clustering result;

[0032] The abnormality analysis module is used to analyze the abnormal state of the transformer based on the data clustering result to obtain the transformer abnormality analysis result.

[0033] Furthermore, the intelligent optimization algorithm in the model optimization module is one of a decentralized search algorithm, a particle swarm algorithm and a northern goshawk algorithm.

[0034] Further, the model optimization module includes an initialization unit, an iterative update unit and a model optimization unit;

[0035] The initialization unit is used to initialize the positions of all northern goshawks in the northern goshawk algorithm to obtain a population matrix of the northern goshawk; the positions of the northern goshawks represent a set of field radii and a minimum number of points contained in the field;

[0036] The iterative updating unit is used to divide the hunting process of the northern goshawk into a first stage optimization search and a second stage optimization search, and iteratively update the position of the northern goshawk based on the first stage optimization search, the second stage optimization search and a preset number of iterations to obtain a target position;

[0037] The model optimization unit is used to determine the domain radius and the minimum number of points contained in the domain in the density clustering optimization model based on the target position, and obtain the density clustering optimization model after parameter optimization.

[0038] Furthermore, the iterative updating unit is also used for:

[0039] The first stage optimization represents the recognition and attack of the northern goshawk on the prey, including updating the position of the northern goshawk based on the fitness function in the density clustering model to obtain a first updated position;

[0040] The second stage optimization represents the northern goshawk chasing the escaped prey, including updating the first updated position based on the goshawk's attack radius to obtain a second updated position;

[0041] The first stage optimization search and the second stage optimization search are repeatedly executed based on the preset number of iterations, and the first updated position and the second updated position are iteratively updated to obtain the target position.

[0042] Furthermore, the clustering module is also used for:

[0043] Traversing each data point in the transformer monitoring data set, and determining the number of data in the field corresponding to each data point based on the optimized field radius;

[0044] Determine the data points whose number of data in the field is greater than the minimum number of points included in the field after optimization as the core objects in the transformer monitoring data set;

[0045] The data points that are density-reachable with each of the core objects are determined recursively, and the density-reachable data points and the corresponding core objects are divided into the same cluster to obtain the data clustering result.

[0046] Furthermore, the abnormality analysis module is also used to:

[0047] Based on the data clustering result, data points in the transformer monitoring data set that do not belong to any cluster are marked to obtain marked abnormal points;

[0048] Sorting the marked abnormal points according to sampling time to obtain an abnormal data sequence;

[0049] If the abnormal data sequence contains n consecutive sampling points, it is determined that the transformer is in an abnormal state.

[0050] Furthermore, the model optimization module is also used for:

[0051] Adopting a plurality of the intelligent optimization algorithms to respectively optimize the parameters in the density clustering model, and obtaining the domain radius and the minimum number of points contained in the domain corresponding to each of the intelligent optimization algorithms;

[0052] Calculate the average value of the domain radius and the average value of the minimum number of included points in the domain based on the domain radius and the minimum number of included points in the domain corresponding to each of the intelligent optimization algorithms;

[0053] The density clustering optimization model is obtained based on the average value of the domain radius and the average value of the minimum number of included points in the domain.

[0054] According to another aspect of the present invention, a storage medium is provided, wherein the storage medium stores at least one executable instruction, and the executable instruction enables a processor to execute operations corresponding to the above transformer abnormality detection method.

[0055] According to another aspect of the present invention, there is provided a computer device, comprising a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus;

[0056] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the above-mentioned transformer abnormality detection method.

[0057] By means of the above technical solution, the technical solution provided by the embodiment of the present invention has at least the following advantages:

[0058] The present invention provides a transformer abnormality detection method and device, storage medium, and computer equipment. Compared with the prior art, the present invention optimizes the domain radius and the minimum number of points contained in the domain in the density clustering model by using an intelligent optimization algorithm to obtain a density clustering optimization model; obtains a transformer monitoring data set, and clusters the transformer monitoring data set using the density clustering optimization model to obtain a data clustering result; analyzes the abnormal state of the transformer based on the data clustering result to obtain a transformer abnormality analysis result, thereby realizing timely analysis of the abnormal state of the transformer. The present invention optimizes and improves the density clustering model by using an intelligent optimization algorithm, so that the clustering effect of the density clustering optimization model is better, and it is easier to improve the accuracy of transformer abnormality detection.

[0059] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0061] Figure 1 A schematic diagram of a process flow of a transformer abnormality detection method provided by an embodiment of the present invention is shown;

[0062] Figure 2 A schematic diagram of a flow chart of another transformer abnormality detection method provided by an embodiment of the present invention is shown;

[0063] Figure 3 A schematic diagram of a flow chart of another transformer abnormality detection method provided by an embodiment of the present invention is shown;

[0064] Figure 4 A schematic diagram of a flow chart of another transformer abnormality detection method provided by an embodiment of the present invention is shown;

[0065] Figure 5 The clustering result based on the SS-DBSCAN model provided by the embodiment of the present invention is shown;

[0066] Figure 6 The clustering result based on the PSO-DBSCAN model provided by the embodiment of the present invention is shown;

[0067] Figure 7The clustering result based on the NGO-DBSCAN model provided by the embodiment of the present invention is shown;

[0068] Figure 8 A schematic diagram of the structure of a transformer abnormality detection device provided by an embodiment of the present invention is shown;

[0069] Fig. 9 A schematic structural diagram of a computer device provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0070] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0071] The embodiment of the present invention provides a transformer abnormality detection method, such as Figure 1 As shown, the method includes:

[0072] 101. An intelligent optimization algorithm is used to optimize the domain radius and the minimum number of points in the domain in the density clustering model to obtain a density clustering optimization model;

[0073] In the embodiment of the present invention, the current execution end uses an intelligent optimization algorithm to optimize the domain radius and the minimum number of points contained in the domain in the density clustering model. Among them, the intelligent optimization algorithm, also known as the modern heuristic algorithm, is an algorithm with global optimization performance, strong versatility and suitable for parallel processing, including: a scattered search algorithm, a genetic algorithm, a simulated annealing algorithm, a taboo search algorithm, a particle swarm algorithm, an ant colony algorithm and a northern goshawk algorithm, etc., which are not specifically limited in the embodiment of the present invention. In the embodiment of the present invention, one of the scattered search algorithm, the particle swarm algorithm and the northern goshawk algorithm is preferred. Among them, the density clustering model (Density-Based Spatial Clustering of Applications with Noise, referred to as DBSCAN) is a relatively representative density-based clustering algorithm. Unlike the partitioning and hierarchical clustering methods, it defines a cluster as the maximum set of points connected by density, can divide areas with sufficiently high density into clusters, and can find clusters of any shape in the spatial database of noise. In the embodiment of the present invention, the current execution end uses an intelligent optimization algorithm to optimize and improve the density clustering model, so that the obtained density clustering optimization model has a better clustering effect, which can effectively improve the accuracy of transformer abnormality analysis.

[0074] 102. Obtain a transformer monitoring data set, and use the density clustering optimization model to perform clustering processing on the transformer monitoring data set to obtain a data clustering result;

[0075] In an embodiment of the present invention, the current execution end obtains a transformer monitoring data set, and the transformer monitoring data set includes data such as the oil temperature, load, and content of various dissolved gases in the transformer oil of the transformer, which is not specifically limited in the embodiment of the present invention. After obtaining the transformer monitoring data set, the current execution end uses the density clustering optimization model obtained by the optimization and improvement of step 101 to cluster the transformer monitoring data set. The clustering process is a typical unsupervised anomaly detection method, which performs clustering based on the density of the data. The core idea is to regard each core object without a category as a seed, and then find a sample set that is density-reachable with it. The largest set of data points connected by a group of densities is called a cluster cluster, thereby obtaining a data clustering result.

[0076] 103. Analyze the abnormal state of the transformer based on the data clustering result to obtain a transformer abnormality analysis result.

[0077] In an embodiment of the present invention, the current execution end analyzes the abnormal state of the transformer based on the data clustering result. The analysis process starts with the generation principle and phenomenon of the transformer abnormal data, analyzes whether the transformer is in an abnormal state from the clustering result, and obtains the transformer abnormality analysis result.

[0078] Further, as a refinement and extension of the specific implementation of the above embodiment, in order to make the density clustering model have a better clustering effect on data, another transformer anomaly detection method is provided, such as Figure 2 As shown, when the intelligent optimization algorithm in the method is the northern goshawk algorithm, the step uses the intelligent optimization algorithm to optimize the domain radius and the minimum number of points contained in the domain in the density clustering model to obtain a density clustering optimization model, including:

[0079] 201. Initialize the positions of all northern goshawks in the northern goshawk algorithm to obtain a population matrix of northern goshawks; the positions of the northern goshawks represent a set of territory radius and a minimum number of points contained in the territory;

[0080] In the embodiment of the present invention, the current execution end needs to initialize the positions of all northern goshawks in the northern goshawk algorithm before optimizing the problem, and obtain the population matrix of the northern goshawk, as shown below:

[0081]

[0082] Where X is the population matrix of the northern goshawk, which contains the positions of all northern goshawks in the search space. Each row represents a position, which is multidimensional so there are several numbers. n represents the population of the northern goshawk, and d represents the dimension of the problem to be solved. Indicates the position of the nth northern goshawk in the dth dimension.

[0083] The positions of northern goshawks in the population matrix of northern goshawks are randomly initialized in the search area and are created as follows:

[0084]

[0085] in, is the position of the i-th northern goshawk in the j-th dimension; UB j , LB j are the upper and lower boundaries of the j-th dimension search domain; t is a random number between (0,1).

[0086] It should be noted that the current execution end in the embodiment of the present invention optimizes the domain radius and the minimum number of points contained in the domain in the density clustering model, so the position of the northern goshawk in the population matrix of the northern goshawk represents a set of domain radius and the minimum number of points contained in the domain; when initializing the upper and lower boundaries of the population, it is necessary to set them according to the actual situation of the domain radius and the minimum number of points contained in the domain. For example, taking the domain radius as an example, the lower boundary of the search domain is set to 0, the upper boundary of the search domain is set to 200, etc., and the embodiment of the present invention does not make specific limitations.

[0087] 202. Divide the hunting process of the northern goshawk into a first stage optimization search and a second stage optimization search, and iteratively update the position of the northern goshawk based on the first stage optimization search, the second stage optimization search and a preset number of iterations to obtain a target position;

[0088] In an embodiment of the present invention, the current execution end divides the hunting process of the northern goshawk into the first stage optimization search and the second stage optimization search, and the first stage optimization search and the second stage optimization search can be regarded as two processes of detection and development. In the first stage optimization search, the northern goshawk will randomly select a prey and quickly attack it. When the northern goshawk hunts the prey, the prey may escape. When the prey escapes, the northern goshawk will enter the second stage of hunting, that is, in the second stage optimization search, the northern goshawk will quickly chase the prey. Under the high-speed movement of the northern goshawk, the prey will eventually be captured by it. This hunting activity is close to an attack position with a radius of R. The current execution end iteratively updates the position of the northern goshawk based on the preset number of iterations T and the divided first stage optimization search and the second stage optimization search to obtain the target position. In an embodiment of the present invention, the target position is the position of the northern goshawk corresponding to the fitness function of the density clustering model when the fitness function obtains the optimal fitness value during the iteration process.

[0089] 203. Determine the domain radius and the minimum number of points contained in the domain in the density clustering optimization model based on the target position, and obtain the density clustering optimization model after parameter optimization.

[0090] In the embodiment of the present invention, the current execution end determines two main parameters (i.e., domain radius and the minimum number of points contained in the domain) in the density clustering optimization model based on the target position, and obtains the density clustering optimization model after parameter optimization. In the density clustering optimization model, the domain radius is used to define the domain range of the data point, denoted by E EPS The minimum number of points in the domain specifies the minimum number of data points that must be included in the domain of a data point, denoted by M. MinPts . The selection of the two parameters, domain radius and the minimum number of points contained in the domain, plays an absolute role in the quality of the clustering results, and the density clustering model DBSCAN is very sensitive to the above two parameters. Any slight difference will lead to very different results. Relying on subjective experience to determine the above two parameters may seriously affect the reliability and accuracy of the density clustering model DBSCAN. This embodiment optimizes the domain radius and the minimum number of points contained in the domain in the density clustering model DBSCAN, which can effectively improve the reliability and accuracy of the density clustering model DBSCAN clustering.

[0091] Furthermore, as a refinement and extension of the specific implementation of the above embodiment, in order to further refine the optimization process of the density clustering model, another transformer anomaly detection method is provided, such as Figure 3 As shown, the step of iteratively updating the position of the northern goshawk based on the first stage optimization, the second stage optimization and a preset number of iterations to obtain a target position includes:

[0092] The first stage optimization represents the recognition and attack of the northern goshawk on the prey, including updating the position of the northern goshawk based on the fitness function in the density clustering model to obtain a first updated position;

[0093] In an embodiment of the present invention, the current execution end needs to select the fitness function in the density clustering model before optimizing in the first stage. In an embodiment of the present invention, the fitness function in the density clustering model is also the objective function of the density clustering model, which can be selected from the clustering effect evaluation indicators, and the clustering effect evaluation indicators include three types: silhouette coefficient, CH score and Davidson-Borgin index DBI. In the Northern Goshawk Algorithm NGO, the particle position seeks optimization towards the maximum value of the fitness function. However, the smaller the Davidson-Borgin index DBI indicator, the better the clustering effect, and the extreme value is zero, so the Davidson-Borgin index DBI is not suitable as the fitness function in the embodiment of the present invention. Both the silhouette coefficient and the CH score are higher, representing better clustering effects. Among them, the silhouette coefficient range is [-1,1], and the CH score has no upper limit. In the experiment, it is easy to get tens of millions or even hundreds of millions of scores, which is not conducive to the presentation of results. In summary, the silhouette coefficient is selected as the fitness function of the model in the embodiment of the present invention. The silhouette coefficient formula is as follows:

[0094] s(i)=(b(i)-a(i)) / max(a(i),b(i))

[0095] Among them, a(i) is the average distance from each data point i in the cluster to all other points in the same cluster, and b(i) is the minimum value of the average distance from data point i to all data points in the cluster that does not contain it.

[0096] In the embodiment of the present invention, the current execution end uses the following formula to represent the behavior of the northern goshawk in the first stage of optimization:

[0097] p i =X k ,k=1,2,...,i-1,i,i+1,...,n

[0098]

[0099] Among them, p i is the location of the prey selected by the i-th northern goshawk, x ij is the position of the population at the current iteration; F i is the fitness value of the population at the current iteration; is the new position of the population after iterative update (where i = 1 or 2); p ij is a random individual different from the current individual; F pi is the fitness value corresponding to the random individual; r2 and r3 are random numbers between (0, 1); I is a random integer of 1 or 2.

[0100] After the population position is updated, whether to adopt the updated population position is determined based on the fitness values ​​before and after the update. The expression is as follows:

[0101]

[0102] The above formula indicates that by comparing the fitness values ​​before and after, it is determined whether the updated position is better than the original position. If it is better, the position is updated, otherwise it is not updated.

[0103] The second stage optimization represents the northern goshawk chasing the escaped prey, including updating the first updated position based on the goshawk's attack radius to obtain a second updated position;

[0104] In the embodiment of the present invention, the current execution end uses the following formula to represent the behavior of the northern goshawk in the second stage of optimization:

[0105]

[0106] Among them, x ij is the population position at the current iteration; is the new position of the population after iterative update (where i=1 or 2); r4 is a random number between (0, 1); t is the current number of iterations; T is the maximum number of iterations. Similar to the first stage optimization, the expression for determining whether to update the population position in the second stage optimization is as follows:

[0107]

[0108] The above formula indicates that by comparing the fitness values ​​before and after, it is determined whether the updated position is better than the original position. If it is better, the position is updated, otherwise it is not updated.

[0109] In the embodiment of the present invention, the current execution end repeatedly executes the first stage optimization search and the second stage optimization search based on the preset number of iterations T, and performs iterative update processing on the first update position and the second update position. If the maximum number of iterations T has been reached, the optimization search ends and the target position is obtained, that is, the optimal global variable E is obtained. EPS With M MinPts .

[0110] Furthermore, as a refinement and extension of the specific implementation of the above embodiment, in order to further refine the clustering process of the density clustering optimization model, another transformer anomaly detection method is provided, such as Figure 3 As shown, the step uses the density clustering optimization model to cluster the transformer monitoring data set to obtain a data clustering result, including:

[0111] 301. Traverse each data point in the transformer monitoring data set, and determine the number of data in the field corresponding to each data point based on the optimized field radius;

[0112] In the embodiment of the present invention, the current execution end combines the transformer monitoring data set and the optimal global variable E in the target position EPS With M MinPts The input is then processed into the density clustering optimization model. The density clustering optimization model needs to traverse every data point in the transformer monitoring data set and calculate the optimal area radius E based on the optimized area radius. EPS Determine the number of data points in the field that correspond to each data point.

[0113] 302. Determine the data points whose number of data in the field is greater than the minimum number of points included in the field after optimization as the core objects in the transformer monitoring data set;

[0114] In the embodiment of the present invention, the current execution end determines multiple core objects from the transformer monitoring data set. The process of determining the core objects is as follows: the number of data in the field corresponding to each data point obtained in step 301 is compared with the minimum number of points M included in the optimized field. MinPts Compare, if the number of data in the field is greater than the minimum number of points M in the field MinPts , the corresponding data point is determined as a core object; otherwise, the corresponding data point is determined as an edge point.

[0115] 303. Determine the data points that are density-reachable with each of the core objects in a recursive manner, and classify the density-reachable data points and the corresponding core objects into the same cluster to obtain the data clustering result.

[0116] In an embodiment of the present invention, the current execution end recursively determines the data points that are density-reachable with each of the core objects. These density-reachable data points are data points connected to the core objects through a series of fields and belong to the same cluster. Specifically, by calculating the distance between the data points, it is determined whether the data points are density-reachable. Then, the current execution end divides the density-reachable data points and the corresponding core objects into the same cluster. When all the data points in the transformer monitoring data set are traversed, the data clustering result is obtained.

[0117] Further, as a refinement and extension of the specific implementation of the above embodiment, in order to timely and effectively judge the abnormal state of the transformer based on the abnormal data in the transformer monitoring data set, another transformer abnormality detection method is provided, such as Figure 4 As shown, the step analyzes the abnormal state of the transformer based on the data clustering result to obtain the transformer abnormality analysis result, including:

[0118] 401. Based on the data clustering result, data points in the transformer monitoring data set that do not belong to any cluster are marked to obtain marked abnormal points;

[0119] In the embodiment of the present invention, the current execution end performs a marking process on the data points in the transformer monitoring data set that do not belong to any cluster based on the data clustering result to obtain a marked abnormal point. For example, the data points belonging to any cluster are marked as 1, and the data points that do not belong to any cluster are marked as 0, and the data points marked as 0 are the marked abnormal points, etc., which are not specifically limited in the embodiment of the present invention.

[0120] 402. Sort the marked abnormal points according to sampling time to obtain an abnormal data sequence;

[0121] 403. If the abnormal data sequence includes n consecutive sampling points, it is determined that the transformer is in an abnormal state.

[0122] In an embodiment of the present invention, the current execution end sorts the marked abnormal points according to the sampling time to obtain an abnormal data sequence. Since the transformer data anomaly is composed of a variety of situations, such as noise interference, data loss, equipment failure, etc. These faults can be divided into two categories, one is abnormal data caused by transformer abnormalities, and the other is abnormal data caused by non-transformer abnormalities. The specific judgment can be based on the time dimension. The abnormal data caused by non-transformer abnormalities such as noise or data loss has a discontinuous feature, while the abnormal data caused by transformer abnormalities has a continuous feature. Therefore, if continuous abnormal data appears, it is judged that the transformer is already in an abnormal state. Therefore, in an embodiment of the present invention, the current execution end analyzes the continuity of the abnormal data in the abnormal data sequence. If the abnormal data sequence contains n consecutive sampling points, it is judged that the transformer is in an abnormal state. Among them, n is an integer, which can be set as needed, and is not specifically limited in the embodiment of the present invention.

[0123] In an embodiment of the present invention, the current execution end takes the hydrogen content dissolved in the oil of a transformer, the oil temperature and the load data as an example, and respectively uses the scattered search algorithm (SS), the particle swarm algorithm (PSO) and the northern goshawk algorithm (NGO) to optimize the density clustering model DBSCAN for parameter optimization and improvement processing, and obtains the density clustering optimization models, which are respectively recorded as SS-DBSCAN, PSO-DBSCAN and NGO-DBSCAN. Then, the current execution end performs clustering processing on the transformer monitoring data set based on the improved density clustering optimization model, and uses the silhouette coefficient, CHI index and DBI index to compare the clustering effect, as shown in Table 1 below:

[0124] Table 1 Clustering effect comparison table

[0125] Clustering method Silhouette Coefficient CHI Index DBI Index SS-DBSCAN 0.7520 1148.3099 0.5798 PSO-DBSCAN 0.7866 1473.2507 0.5147 NGO-DBSCAN 0.8281 1870.2133 0.4487

[0126] From the comparison results in Table 1, we can see that the clustering effect of NGO-DBSCAN is better than that of the other two models.

[0127] In addition, the current execution end also marks the abnormal data based on the clustering results of the above-mentioned SS-DBSCAN model, PSO-DBSCAN model and NGO-DBSCAN model, and sorts the marked transformer monitoring data set according to the sampling time to obtain the following: Figure 5 , Figure 6 , Figure 7 The abnormal data sequence shown. Figure 5 It can be seen from the figure that the SS-DBSACN model has continuous abnormal data since T=405, and it is judged that the transformer has an abnormal state; Figure 6 It can be seen from the figure that the PSO-DBSCAN model has continuous abnormal data since T=392, and it is judged that the transformer has an abnormal state; Figure 7 It can be seen that the NGO-DBSCAN model has continuous abnormal data since T=379, and it is judged that the transformer has an abnormal state. It can be seen that the NGO-DBSCAN model can detect the abnormality of the transformer earlier and detect abnormal data more accurately.

[0128] Further, as a refinement and extension of the specific implementation of the above embodiment, in order to optimize the parameters in the density clustering model in combination with the optimization results of each intelligent optimization algorithm, another transformer abnormality detection method is provided, wherein the steps use an intelligent optimization algorithm to optimize the domain radius and the minimum number of points contained in the domain in the density clustering model to obtain a density clustering optimization model, including:

[0129] Adopting a plurality of the intelligent optimization algorithms to respectively optimize the parameters in the density clustering model, and obtaining the domain radius and the minimum number of points contained in the domain corresponding to each of the intelligent optimization algorithms;

[0130] Calculate the average value of the domain radius and the average value of the minimum number of included points in the domain based on the domain radius and the minimum number of included points in the domain corresponding to each of the intelligent optimization algorithms;

[0131] The density clustering optimization model is obtained based on the average value of the domain radius and the average value of the minimum number of included points in the domain.

[0132] The embodiment of the present invention provides a transformer abnormality detection method. Compared with the prior art, the present invention optimizes the domain radius and the minimum number of points contained in the domain in the density clustering model by using an intelligent optimization algorithm to obtain a density clustering optimization model; obtains a transformer monitoring data set, and clusters the transformer monitoring data set using the density clustering optimization model to obtain a data clustering result; analyzes the abnormal state of the transformer based on the data clustering result to obtain a transformer abnormality analysis result, thereby realizing timely analysis of the abnormal state of the transformer. The present invention optimizes and improves the density clustering model by using an intelligent optimization algorithm, so that the clustering effect of the density clustering optimization model is better, and it is easier to improve the accuracy of transformer abnormality detection.

[0133] As the above Figure 1 The embodiment of the present invention provides a transformer abnormality detection device, such as Figure 8 As shown, the device comprises:

[0134] The model optimization module 51 is used to optimize the domain radius and the minimum number of points in the domain in the density clustering model by using an intelligent optimization algorithm to obtain a density clustering optimization model;

[0135] A clustering module 52 is used to obtain a transformer monitoring data set, and perform clustering processing on the transformer monitoring data set using the density clustering optimization model to obtain a data clustering result;

[0136] The abnormality analysis module 53 is used to analyze the abnormal state of the transformer based on the data clustering result to obtain the transformer abnormality analysis result.

[0137] Furthermore, the intelligent optimization algorithm in the model optimization module 51 is one of a decentralized search algorithm, a particle swarm algorithm and a northern goshawk algorithm.

[0138] Further, the model optimization module 51 includes an initialization unit, an iterative update unit and a model optimization unit;

[0139] The initialization unit is used to initialize the positions of all northern goshawks in the northern goshawk algorithm to obtain a population matrix of the northern goshawk; the positions of the northern goshawks represent a set of field radii and a minimum number of points contained in the field;

[0140] The iterative updating unit is used to divide the hunting process of the northern goshawk into a first stage optimization search and a second stage optimization search, and iteratively update the position of the northern goshawk based on the first stage optimization search, the second stage optimization search and a preset number of iterations to obtain a target position;

[0141] The model optimization unit is used to determine the domain radius and the minimum number of points contained in the domain in the density clustering optimization model based on the target position, and obtain the density clustering optimization model after parameter optimization.

[0142] Furthermore, the iterative updating unit is also used for:

[0143] The first stage optimization represents the recognition and attack of the northern goshawk on the prey, including updating the position of the northern goshawk based on the fitness function in the density clustering model to obtain a first updated position;

[0144] The second stage optimization represents the northern goshawk chasing the escaped prey, including updating the first updated position based on the goshawk's attack radius to obtain a second updated position;

[0145] The first stage optimization search and the second stage optimization search are repeatedly executed based on the preset number of iterations, and the first updated position and the second updated position are iteratively updated to obtain the target position.

[0146] Furthermore, the clustering module 52 is also used for:

[0147] Traversing each data point in the transformer monitoring data set, and determining the number of data in the field corresponding to each data point based on the optimized field radius;

[0148] Determine the data points whose number of data in the field is greater than the minimum number of points included in the field after optimization as the core objects in the transformer monitoring data set;

[0149] The data points that are density-reachable with each of the core objects are determined recursively, and the density-reachable data points and the corresponding core objects are divided into the same cluster to obtain the data clustering result.

[0150] Furthermore, the abnormality analysis module 53 is also used for:

[0151] Based on the data clustering result, data points in the transformer monitoring data set that do not belong to any cluster are marked to obtain marked abnormal points;

[0152] Sorting the marked abnormal points according to sampling time to obtain an abnormal data sequence;

[0153] If the abnormal data sequence contains n consecutive sampling points, it is determined that the transformer is in an abnormal state.

[0154] Furthermore, the model optimization module 51 is also used for:

[0155] Adopting a plurality of the intelligent optimization algorithms to respectively optimize the parameters in the density clustering model, and obtaining the domain radius and the minimum number of points contained in the domain corresponding to each of the intelligent optimization algorithms;

[0156] Calculate the average value of the domain radius and the average value of the minimum number of included points in the domain based on the domain radius and the minimum number of included points in the domain corresponding to each of the intelligent optimization algorithms;

[0157] The density clustering optimization model is obtained based on the average value of the domain radius and the average value of the minimum number of included points in the domain.

[0158] The embodiment of the present invention provides a transformer abnormality detection device. Compared with the prior art, the present invention optimizes the domain radius and the minimum number of points contained in the domain in the density clustering model by using an intelligent optimization algorithm to obtain a density clustering optimization model; obtains a transformer monitoring data set, and clusters the transformer monitoring data set using the density clustering optimization model to obtain a data clustering result; analyzes the abnormal state of the transformer based on the data clustering result to obtain a transformer abnormality analysis result, thereby realizing timely analysis of the abnormal state of the transformer. The present invention optimizes and improves the density clustering model by using an intelligent optimization algorithm, so that the clustering effect of the density clustering optimization model is better, and it is easier to improve the accuracy of transformer abnormality detection.

[0159] According to an embodiment of the present invention, a storage medium is provided, wherein the storage medium stores at least one executable instruction, and the computer executable instruction can execute the transformer abnormality detection method in any of the above method embodiments.

[0160] Fig. 9 A schematic diagram of the structure of a computer device provided according to an embodiment of the present invention is shown. The specific embodiment of the present invention does not limit the specific implementation of the computer device.

[0161] like Fig. 9 As shown, the computer device may include: a processor (processor) 602 , a communication interface (Communications Interface) 604 , a memory (memory) 606 , and a communication bus 608 .

[0162] The processor 602 , the communication interface 604 , and the memory 606 communicate with each other via a communication bus 608 .

[0163] The communication interface 604 is used to communicate with other devices such as clients or other servers.

[0164] The processor 602 is used to execute the program 610, and specifically can execute the relevant steps of the above transformer abnormality detection method.

[0165] Specifically, the program 610 may include program codes, which include computer operation instructions.

[0166] The processor 602 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention. The one or more processors included in the computer device may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0167] The memory 606 is used to store the program 610. The memory 606 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0168] The program 610 may be specifically configured to enable the processor 602 to perform the following operations:

[0169] An intelligent optimization algorithm is used to optimize the domain radius and the minimum number of points in the domain in the density clustering model to obtain a density clustering optimization model.

[0170] Acquire a transformer monitoring data set, and use the density clustering optimization model to perform clustering processing on the transformer monitoring data set to obtain a data clustering result;

[0171] The abnormal state of the transformer is analyzed based on the data clustering result to obtain a transformer abnormality analysis result.

[0172] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0173] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A transformer abnormality detection method, characterized in that: include: An intelligent optimization algorithm is used to optimize the domain radius and the minimum number of points in the domain in the density clustering model to obtain a density clustering optimization model. Acquire a transformer monitoring data set, and use the density clustering optimization model to perform clustering processing on the transformer monitoring data set to obtain a data clustering result; The abnormal state of the transformer is analyzed based on the data clustering result to obtain a transformer abnormality analysis result.

2. The method according to claim 1, characterized in that The intelligent optimization algorithm is one of a decentralized search algorithm, a particle swarm algorithm and a northern goshawk algorithm.

3. The method according to claim 2, characterized in that When the intelligent optimization algorithm is the Northern Goshawk algorithm, the intelligent optimization algorithm is used to optimize the domain radius and the minimum number of points contained in the domain in the density clustering model to obtain a density clustering optimization model, including: Initializing the positions of all northern goshawks in the northern goshawk algorithm to obtain a population matrix of the northern goshawk; the positions of the northern goshawks represent a set of field radii and a minimum number of points contained in the field; The hunting process of the northern goshawk is divided into a first stage optimization search and a second stage optimization search, and the position of the northern goshawk is iteratively updated based on the first stage optimization search, the second stage optimization search and a preset number of iterations to obtain a target position; Based on the target position, the domain radius and the minimum number of points contained in the domain in the density clustering optimization model are determined to obtain the density clustering optimization model after parameter optimization.

4. The method according to claim 3, characterized in that The iterative updating process of the position of the northern goshawk based on the first stage optimization search, the second stage optimization search and a preset number of iterations to obtain a target position includes: The first stage optimization represents the recognition and attack of the northern goshawk on the prey, including updating the position of the northern goshawk based on the fitness function in the density clustering model to obtain a first updated position; The second stage optimization represents the northern goshawk chasing the escaped prey, including updating the first updated position based on the goshawk's attack radius to obtain a second updated position; The first stage optimization search and the second stage optimization search are repeatedly executed based on the preset number of iterations, and the first updated position and the second updated position are iteratively updated to obtain the target position.

5. The method according to claim 1, characterized in that The method of clustering the transformer monitoring data set using the density clustering optimization model to obtain a data clustering result includes: Traversing each data point in the transformer monitoring data set, and determining the number of data in the field corresponding to each data point based on the optimized field radius; Determine the data points whose number of data in the field is greater than the minimum number of points included in the field after optimization as the core objects in the transformer monitoring data set; The data points that are density-reachable with each of the core objects are determined recursively, and the density-reachable data points and the corresponding core objects are divided into the same cluster, so as to obtain the data clustering result.

6. The method according to claim 1, characterized in that The abnormal state of the transformer is analyzed based on the data clustering result to obtain the transformer abnormality analysis result, including: Based on the data clustering result, data points in the transformer monitoring data set that do not belong to any cluster are marked to obtain marked abnormal points; Sorting the marked abnormal points according to sampling time to obtain an abnormal data sequence; If the abnormal data sequence contains n consecutive sampling points, it is determined that the transformer is in an abnormal state, where n is a positive integer.

7. The method according to any one of claims 1, 5 and 6, characterized in that The intelligent optimization algorithm is used to optimize the domain radius and the minimum number of points in the domain in the density clustering model to obtain a density clustering optimization model, including: Adopting a plurality of the intelligent optimization algorithms to respectively optimize the parameters in the density clustering model, and obtaining the domain radius and the minimum number of points contained in the domain corresponding to each of the intelligent optimization algorithms; Calculate the average value of the domain radius and the average value of the minimum number of included points in the domain based on the domain radius and the minimum number of included points in the domain corresponding to each of the intelligent optimization algorithms; The density clustering optimization model is obtained based on the average value of the domain radius and the average value of the minimum number of included points in the domain.

8. A transformer abnormality detection device, characterized in that: include: The model optimization module is used to optimize the domain radius and the minimum number of points in the domain in the density clustering model using an intelligent optimization algorithm to obtain a density clustering optimization model; A clustering module, used to obtain a transformer monitoring data set, and perform clustering processing on the transformer monitoring data set using the density clustering optimization model to obtain a data clustering result; The abnormality analysis module is used to analyze the abnormal state of the transformer based on the data clustering result to obtain the transformer abnormality analysis result.

9. A storage medium, wherein at least one executable instruction is stored in the storage medium, and the executable instruction executes an operation corresponding to the transformer abnormality detection method according to any one of claims 1 to 7.

10. A computer device, comprising a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the transformer abnormality detection method according to any one of claims 1-7.

Citation Information

Patent Citations

  • Transformer state data cleaning method and device, storage medium and computer equipment

    CN117493761A

  • Method and system for abnormal detection of power monitoring operation data

    CN118152962A