A method and system for locating equipment defects of a GIS device

Abstract

The present invention relates to the technical field of defect detection, and discloses a method and system for locating equipment defects of a GIS device, which are used to improve the accuracy of locating equipment defects of the GIS device. The method includes: collecting the original operation data of the preset GIS device, and converting the original operation data into fused multi-source data; performing multi-scale decomposition on the fused multi-source data to obtain multi-scale decomposition data, and performing sparse representation processing on the multi-scale decomposition data to obtain sparse representation data; performing dynamic time warping processing on the sparse representation data to obtain time series feature data; extracting fractal dimension features from the time series feature data to obtain fractal feature data; performing spectral clustering processing on the fractal feature data to obtain clustering feature data; inputting the clustering feature data into a local outlier factor analysis algorithm for outlier extraction to obtain outliers, and determining equipment defect location data according to the outliers.

Description

Technical Field

[0001] The present invention relates to the technical field of defect detection, and particularly to a method and system for locating equipment defects of a GIS device. Background Art

[0002] A GIS device (gas insulated switchgear) is a highly efficient and reliable power equipment, which is widely used in power transmission and distribution systems. To ensure the normal operation of the GIS device and extend its service life, conventional maintenance methods include regular inspections and fault diagnosis. Currently, the fault diagnosis methods of GIS devices mainly rely on manual inspections and simple monitoring devices, such as temperature sensors and pressure sensors. These methods can detect abnormal conditions of the equipment to a certain extent, but there are often problems of low detection accuracy and poor real-time performance, and the specific defect locations of the equipment cannot be accurately located in a timely manner.

[0003] The existing fault diagnosis methods of GIS devices have the following deficiencies. First, it is difficult for manual inspections and simple monitoring devices to capture subtle changes during equipment operation, and potential fault hazards are easily missed. Second, the existing methods rely on expert experience, are difficult to standardize and automate, and it is difficult to guarantee the accuracy and consistency of diagnosis results. In addition, most of the existing monitoring devices are single sensors, lacking the ability of multi-source data fusion and in-depth analysis, resulting in inaccurate and incomplete location of equipment defects. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed. The present invention provides a method and system for locating equipment defects of a GIS device, which is used to improve the accuracy rate of locating equipment defects of a GIS device.

[0005] To solve the above technical problems, a method for locating equipment defects of a GIS device is proposed, including,

[0006] Collecting the original operation data of a preset GIS device, and converting the original operation data into multi-source data fusion; performing multi-scale decomposition on the multi-source data fusion to obtain multi-scale decomposition data, and performing sparse representation processing on the multi-scale decomposition data to obtain sparse representation data; performing dynamic time warping processing on the sparse representation data to obtain time series feature data; extracting fractal dimension features from the time series feature data to obtain fractal feature data; performing spectral clustering processing on the fractal feature data to obtain clustering feature data; inputting the clustering feature data into a local outlier factor analysis algorithm for outlier extraction to obtain outliers, and determining equipment defect location data according to the outliers.

[0007] As a preferred solution of a method for locating equipment defects of a GIS device according to the present invention, wherein: the fusion of multi-source data includes collecting the operation data of sensors arranged during the operation of a preset GIS device to obtain original operation data D = {d1, d2, …, d t}, where d t represents the data at the t-th time point.

[0008] The original operation data includes equipment status data, environmental data, and maintenance history data.

[0009] The equipment status data includes temperature, humidity, voltage, current, vibration, sound, insulating gas concentration, and the number of breaker operations.

[0010] The environmental data includes external temperature, precipitation data, and air quality.

[0011] The maintenance history data includes maintenance records, fault history, and equipment structure parameters.

[0012] Perform outlier detection Detect(t) on the original operation data, and set a weighted moving average function WMA(t):

[0013]

[0014] where W is the time window, T is the time for obtaining data, and σ WMA is the standard deviation of the Gaussian kernel, controlling the speed of weight decay.

[0015] Monitor abnormal data I(t) by accumulating abnormal scores within the T time period:

[0016]

[0017]

[0018] where S(t) is the abnormal score data function. When I(t) > θ and θ = 4, let Detect(t) be 1. When I(t) ≤ θ, let Detect(t) be 0. The final detection model is:

[0019] Anomaly(t) = Detect(t) · d t

[0020] When Anomaly(t) = 0, it means that no anomaly is detected at time point t. When Anomaly(t) = d t it means that an anomaly is detected at time point t, and d t is the value of the abnormal data, obtaining the outlier data.

[0021] Classify the outlier data according to features and attributes to obtain data dimensions with different attributes, and perform data fusion on the original operation data based on the data dimensions to obtain fused multi-source data.

[0022] As a preferred solution of the device defect location method for a GIS device according to the present invention, wherein: the sparse representation data includes inputting the fused multi-source data into a preset empirical mode decomposition algorithm to construct a signal envelope, obtaining a signal envelope Env(t), where the signal envelope includes an upper envelope and a lower envelope:

[0023]

[0024] Wherein, EMD is the empirical mode decomposition function for decomposing multi-source data, x and y respectively represent the components of the signal in the horizontal and vertical directions, μ i and σ i are respectively the mean and variance of the Gaussian distribution of the i-th data, representing the continuous part of the signal. λ j and ν j are respectively the rate parameter and location parameter of the exponential distribution of the j-th data, representing the transient part of the signal.

[0025] Calculate the mean values of the upper envelope and the lower envelope to obtain a target mean value:

[0026]

[0027] Wherein, represents the target mean value of the upper and lower envelopes, Env u (t) and Env l (t) respectively represent the upper envelope and the lower envelope, and t1 and t2 are the time intervals for calculating the mean value.

[0028] Perform a first decomposition on the fused multi-source data based on the target mean value to obtain a plurality of first decomposition data, and perform the first decomposition on the fused multi-source data based on the target mean value respectively: subtract the target mean value from the original signal to obtain a new signal, and the new signal is the residual of the original signal after removing the local mean value. When the residual still contains local maxima and minima, then repeat removing the local mean value, continue to perform envelope construction and target mean value calculation until the residual no longer contains significant local extreme values.

[0029] Embed white noise into the obtained first decomposition data to obtain second decomposition data, and perform an averaging process on all the second decomposition data to obtain multi-scale decomposition data.

[0030] Perform sparse coding on the multi-scale decomposition data to obtain coded decomposition data.

[0031] Perform sparse representation processing on the encoded decomposition data to obtain sparse representation data.

[0032] As a preferred solution of a method for locating equipment defects of a GIS device according to the present invention, wherein: the dynamic time warping processing includes dividing the sparse representation data into different data points, each data point representing the equipment state at a moment, performing data point matching, and respectively analyzing the distances between adjacent data points for each data point to obtain the distance data f1(X a ,X b ):

[0033]

[0034] Calculate the cumulative distance matrix F for each data point based on the distance data corresponding to each data point ab :

[0035]

[0036] Generate the backtracking path L for each data point through the cumulative distance matrix of each data point a :

[0037]

[0038] Extract the time series features h(X, t) from the sparse representation data based on the backtracking path of each data point to obtain the time series feature data Y a :

[0039]

[0040] Wherein, X a and X b are different data points obtained by segmentation, representing the state of the GIS device at a specific moment. F ab represents the cumulative distance from the data point X a to the data point X b , X k is the intermediate data point during the accumulation process, L a is the backtracking path of the a-th data point, representing the optimal path from the current data point back to the starting point, g(a) represents the importance of the b-th data point, w is the total number of data points, t a and t a+1 respectively represent the time points corresponding to the data points X a and X a+1 .

[0041] As a preferred solution of a method for locating equipment defects of a GIS device according to the present invention, wherein: the fractal dimension feature extraction includes the time-series feature data Y a Calibrate the data range to obtain the target data range, and based on the target data range, segment the time-series feature data to obtain different time-series feature data subsequences B with the same time interval T: B1{Y a1 , Y a2 , … Y ai}, B2{Y a(i+1) , Y a(i+2) , … Y a(i+j )} … B n {Y a(i+j+1) , Y a(i+j+2) , … Y an}, and assign measurement values to each subsequence respectively. Among them, the measurement value is the energy fluctuation intensity, which is obtained by calculating the energy of the subsequence E = {Y a1 2 + Y a2 2 + … Y ai 2}, {Y a(i+1) 2 + Y a(i+2) 2 + … + Y a(i+j) 2} … {Y a(i+j+1) 2 + Y a(i+j+2) 2 + … + Y an 2}.

[0042] Calculate the fractal dimension of each subsequence through the box dimension and multifractal spectrum methods based on the measurement values of each subsequence to obtain the fractal dimension G(B) of each subsequence:

[0043]

[0044] Among them, A is the length of the subsequence, N is the number of sample points of the subsequence, and α is the weight for adjusting the measurement value.

[0045] Construct the multifractal spectrum f(γ) according to the fractal dimension of each subsequence:

[0046]

[0047] Among them, p(B[n]) is the normalized probability distribution of the subsequence B, δ is the pulse of the subsequence sample point, and γ is the weight index of the fractal spectrum.

[0048] Feature extraction is performed on the multifractal spectrum to obtain the fractal feature data H:

[0049]

[0050] Wherein, is the peak value of the multifractal spectrum, is the average size of the signal scale change, is the dispersion degree of the signal scale change, and <γ> is the expected value of γ.

[0051] As a preferred solution of the device defect location method for a GIS device according to the present invention, wherein: the spectral clustering process includes calculating the feature similarity similarity(H i ,H j ) of the fractal feature data H:

[0052] similarity(H i ,H j ) = exp(-β||H i -H j )

[0053] A similarity data set is obtained, wherein β is the bandwidth parameter of the Gaussian kernel, controlling the smoothness of the function.

[0054] Construct a Laplacian matrix based on the similarity data set Wherein, C is the degree of each vertex in the Laplacian matrix, L norm is the unnormalized Laplacian matrix, and the Laplacian matrix is normalized to obtain a normalized Laplacian matrix and eigenvalue decomposition is performed to obtain multiple decomposition eigenvalues Wherein, V is the eigenvector, is the corresponding eigenvalue.

[0055] Perform spectral clustering processing on the decomposition eigenvalues to obtain clustering feature data K:

[0056]

[0057]

[0058] Wherein, is the eigenvector matrix composed of the eigenvectors corresponding to the first m smallest eigenvalues, V m is the m-th eigenvector, S i is the i-th cluster.

[0059] As a preferred solution of the device defect location method for a GIS device according to the present invention, wherein: determining the device defect location data includes inputting the clustering feature data K into a local outlier factor analysis algorithm for local density analysis to obtain local density data ρ(K):

[0060]

[0061] where Ω is the spatial region of data points, ‖K a -K b ‖ is the Euclidean distance between point K a and K b , ε is the width parameter of the kernel function, and f(K b ) is the clustering feature of point K b .

[0062] Match the local density data with neighborhood density data to obtain neighborhood density data ω(K):

[0063]

[0064] where P is the number of data points in the neighborhood, ψ is the parameter controlling the curvature of the function, and r(K, K a ) is used to compare the relationship between two data points.

[0065] Calculate the reachability distance η(K) of the data points through the neighborhood density data:

[0066]

[0067] where τ and ξ are the parameters controlling the shape and scale of the function.

[0068] Extract outlier points from the clustering feature data through the reachability distance of the data points to obtain outlier points, and determine the device defect location data according to the outlier points: when the local density data ρ(K) of a data point is higher than 2.5, it means that the local density of the current data point is lower than the neighborhood density.

[0069] Another object of the present invention is to provide a device defect location system for GIS devices. The present invention realizes the comprehensive monitoring of the operation state of GIS devices and fault early warning, and improves the safety and economy of the power system. The system first obtains the original operation data of GIS devices through the acquisition module and converts it into multi-source data fusion. The decomposition module performs multi-scale decomposition and sparse representation processing on the multi-source data to effectively extract the feature information of the device at different scales. The regularization module processes the data using the dynamic time warping technique to eliminate the fluctuations in the time series and enhance the comparability of the features. The extraction module further extracts the fractal dimension features of the time series data to reveal the inherent complexity and self-similarity of the data. The clustering module processes the fractal feature data through the spectral clustering method to classify similar feature data for subsequent analysis. Finally, the location module combines the local outlier factor analysis algorithm to identify outliers from the clustering feature data, thereby accurately locating the device defects. Through the collaborative work of these modules, the system solves the problem that it is difficult to accurately detect and locate the defects of GIS devices by traditional methods, realizes the comprehensive monitoring of the operation state of GIS devices and fault early warning, and improves the safety and economy of the power system.

[0070] As a preferred solution of a device defect location system for GIS devices according to the present invention, it is characterized in that it includes an acquisition module, a decomposition module, a regularization module, an extraction module, a clustering module, and a location module.

[0071] The acquisition module is used to collect the original operation data of the preset GIS devices and convert the original operation data into multi-source data fusion.

[0072] The decomposition module is used to perform multi-scale decomposition on the multi-source data fusion to obtain multi-scale decomposition data, and perform sparse representation processing on the multi-scale decomposition data to obtain sparse representation data.

[0073] The regularization module is used to perform dynamic time warping processing on the sparse representation data to obtain time series feature data.

[0074] The extraction module is used to extract the fractal dimension features from the time series feature data to obtain fractal feature data.

[0075] The clustering module is used to perform spectral clustering processing on the fractal feature data to obtain clustering feature data.

[0076] The location module is used to input the clustering feature data into the local outlier factor analysis algorithm for outlier extraction to obtain outliers, and determine the device defect location data according to the outliers.

[0077] A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that when the processor executes the computer program, the steps of a method for locating device defects of a GIS device are implemented.

[0078] 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 method for locating device defects of a GIS device are implemented.

[0079] Advantages of the present invention: In the technical solution provided by the present invention, through a series of steps such as multi-source data fusion, multi-scale decomposition, sparse representation, dynamic time warping, fractal dimension feature extraction, spectral clustering, and local outlier factor analysis, precise positioning of GIS equipment defects is achieved, with significant beneficial effects. First, by collecting the original operation data of GIS equipment and performing multi-source data fusion, the comprehensiveness and diversity of the data are ensured, providing a reliable data basis for subsequent analysis and processing. Multi-source data fusion can effectively integrate data from different sensors, make up for the deficiencies of single-sensor data, and improve the richness and accuracy of the data. Secondly, in the multi-scale decomposition process, empirical mode decomposition (EMD) and complete ensemble empirical mode decomposition (CEEMD) methods are adopted. These methods can adaptively decompose non-linear and non-stationary signals, extract signal features at different time scales, and ensure the accuracy and meticulousness of the decomposition results. At the same time, through sparse representation processing, high-dimensional data is converted into low-dimensional sparse representation, effectively reducing the data dimension and improving the computational efficiency of subsequent processing and the robustness of data processing. The dynamic time warping (DTW) processing step calculates the similarity of signals in different time periods in the sparse representation data, finds the optimal alignment path, and extracts time series feature data. This step can effectively align signals with time offsets, extract key features reflecting the dynamic changes of signals, and enhance the accuracy and consistency of time series analysis. In the fractal dimension feature extraction process, by combining the box dimension and multi-fractal spectrum methods, the fractal dimension features of the time series feature data are calculated to obtain more detailed and diverse fractal feature data. The fractal dimension can reveal the internal complexity and multi-scale characteristics of the data, providing rich feature information for subsequent clustering analysis. The spectral clustering processing step maps high-dimensional data to a low-dimensional space through feature similarity calculation, construction and normalization of the Laplacian matrix, and eigenvalue decomposition method for clustering analysis to obtain clustering feature data. This method can effectively identify different patterns and structures in the data, improving the accuracy and stability of the clustering results. Finally, through local outlier factor (LOF) analysis, local density analysis and outlier detection are performed on the clustering feature data to identify outliers with significantly lower local density than the neighborhood, and finally determine the defect location data of the equipment. The LOF algorithm can effectively detect local outliers within the cluster, further improving the accuracy and reliability of defect positioning. Description of the Drawings

[0080] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings, where:

[0081] Figure 1 The overall flowchart of a method for locating device defects of a GIS device provided by an embodiment of the present invention.

[0082] Figure 2 The system functional architecture diagram of a system for locating device defects of a GIS device provided by an embodiment of the present invention. Detailed implementation manners

[0083] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe the detailed implementation manners of the present invention in conjunction with the accompanying drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0084] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0085] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it an embodiment that is separately or selectively mutually exclusive with other embodiments.

[0086] The present invention is described in detail in conjunction with schematic diagrams. When describing the embodiments of the present invention in detail, for the convenience of explanation, the cross-sectional views showing the device structure will be enlarged locally not in accordance with the general scale, and the schematic diagrams are only examples and should not limit the protection scope of the present invention here. In addition, in actual production, three-dimensional spatial dimensions including length, width, and depth should be included.

[0087] At the same time, in the description of the present invention, it should be noted that the orientation or positional relationship indicated by terms such as "upper, lower, inner, and outer" is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation to the present invention. In addition, the terms "first, second, or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0088] Unless otherwise clearly defined and limited in the present invention, the terms "installation, connection, and coupling" shall be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may also be a mechanical connection, an electrical connection, or a direct connection, or may be indirectly connected through an intermediate medium, or may be the communication inside two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0089] Example 1

[0090] Referring to Figure 1 , which is the first embodiment of the present invention. This embodiment provides a method for locating equipment defects of a GIS device, including:

[0091] S1: Collect the original operation data of a preset GIS device, and convert the original operation data into fused multi-source data.

[0092] Furthermore, collect the operation data of sensors arranged during the operation of a preset GIS device to obtain the original operation data D = {d1, d2,..., d t}, where d t represents the data at the t-th time point.

[0093] The original operation data includes equipment status data, environmental data, and maintenance history data.

[0094] The equipment status data includes temperature, humidity, voltage, current, vibration, sound, insulating gas concentration, and the number of breaker operations.

[0095] The environmental data includes external temperature, precipitation data, and air quality.

[0096] The maintenance history data includes maintenance records, fault history, and equipment structure parameters.

[0097] Perform outlier detection Detect(t) on the original operation data, and set the weighted moving average function WMA(t):

[0098]

[0099] where W is the time window, T is the time for obtaining data, and σ WMA is the standard deviation of the Gaussian kernel, which controls the speed of weight decay.

[0100] Monitor the abnormal data I(t) by accumulating the abnormal scores within the T time period:

[0101]

[0102]

[0103] Among them, S(t) is the anomaly score data function. When I(t) > θ, where θ = 4, let Detect(t) be 1. When I(t) ≤ θ, let Detect(t) be 0. The final detection model is:

[0104] Anomaly(t) = Detect(t)·d t

[0105] When Anomaly(t) = 0, it means that no anomaly is detected at time point t. When Anomaly(t) = d t it means that an anomaly is detected at time point t, and d t is the value of the anomaly data, obtaining the anomaly value data.

[0106] Classify the anomaly value data according to features and attributes to obtain data dimensions with different attributes. Based on the data dimensions, perform data fusion on the original operation data to obtain fused multi-source data.

[0107] By analyzing historical data, the operation data of GIS devices usually follows a normal distribution, and 99% of the data is within ±3 standard deviations of the mean. Selecting θ = 4 standard deviations means that only about 0.3% of the data will be considered anomalous, thus reducing the false alarm rate; selecting θ = 4 provides a relatively loose threshold, making the system insensitive to some minor fluctuations, thereby reducing unnecessary alarms.

[0108] It should be noted that during the operation of GIS equipment, various pre-set sensors are used to collect its operation data in real time to obtain the original operation data. These sensors include, but are not limited to, temperature sensors, current sensors, voltage sensors, vibration sensors, and noise sensors, etc. Each sensor records the parameter data of the equipment under different operation states in real time. For example, the temperature sensor can record the temperature changes of each component during the operation of the GIS equipment, the current sensor can record the current fluctuations, and the vibration sensor can capture the amplitude and frequency of the mechanical vibration of the equipment. These sensors transmit the collected original data to the central processing system through the data acquisition system for subsequent data processing and analysis. Next, outlier detection is performed on the collected original operation data with the aim of removing outliers and noise in the data to ensure the accuracy and reliability of the data. Outlier detection uses methods based on time series analysis, such as autoregressive moving average models, to predict the normal data range and detect outlier data points. For example, if a certain voltage sensor records a series of abnormally high voltages in a short period of time, this may be due to sensor failure or external interference. After outlier detection, these data points will be identified and removed. After completing the outlier detection, the detected outlier data is classified by data dimension to obtain multiple data dimensions. The purpose of data dimension classification is to classify different types of data according to their characteristics and attributes for subsequent data fusion processing. Each type of data represents a dimension. For example, temperature data, current data, voltage data, vibration data, and noise data represent different data dimensions respectively. After classification, the data in each data dimension has the same or similar physical meaning and unit, which is convenient for unified processing and analysis. For example, when classifying temperature data, the temperature data of different components can be grouped into the same dimension, while current and voltage data are grouped into different dimensions respectively. Next, based on multiple data dimensions, data fusion is performed on the original operation data to obtain fused multi-source data. The purpose of data fusion is to integrate the data from different sensors to form a unified and comprehensive data set. For example, when processing temperature and voltage data, different weights can be assigned to the temperature and voltage data respectively according to the accuracy of the sensors and the importance of the data, and they are weighted and averaged to obtain a comprehensive data set that integrates temperature and voltage information. To better illustrate this process, a specific example can be given. Suppose a certain GIS equipment is installed with multiple temperature sensors, current sensors, and vibration sensors during operation. The temperature sensors record the temperature changes of different components of the equipment, such as the temperature of the main switch, the temperature of the insulating material, and the temperature of the connection point; the current sensors record the current changes of different circuits, such as the current of the main circuit and the auxiliary circuit; the vibration sensors record the vibration amplitude of the whole and local parts of the equipment. At a certain moment, the temperature sensor records that the temperature of the main switch suddenly rises to an abnormally high value, which may be due to sensor failure or external environmental influence.After outlier detection, this abnormally high temperature point was identified and removed. At the same time, the current sensor recorded that the current fluctuation in the main circuit was abnormally large, and the vibration sensor recorded that the vibration amplitude at a certain part of the equipment was abnormally high. After outlier detection, these data were respectively classified into the current data dimension and the vibration data dimension. After the data dimension classification was completed, through data fusion technology, these multi-dimensional data were fused. For example, using the weighted average method, different weights can be assigned to temperature, current, and vibration data according to the accuracy of each sensor and the importance of the data, and they are weighted and averaged to form a comprehensive fused multi-source data set. In this data set, each data point not only contains temperature information but also synthesizes current and vibration information, reflecting the comprehensive operating state of the equipment at that moment.

[0109] S2: Perform multi-scale decomposition on the fused multi-source data to obtain multi-scale decomposition data, and perform sparse representation processing on the multi-scale decomposition data to obtain sparse representation data.

[0110] Furthermore, input the fused multi-source data into a preset empirical mode decomposition algorithm to construct a signal envelope, obtaining the signal envelope Env(t), where the signal envelope includes an upper envelope and a lower envelope:

[0111]

[0112] Among them, EMD is the empirical mode decomposition function for decomposing multi-source data, x and y respectively represent the components of the signal in the horizontal and vertical directions, μ i and σ i are respectively the mean and variance of the Gaussian distribution of the i-th data, representing the continuous part of the signal. λ j and ν j are respectively the rate parameter and location parameter of the exponential distribution of the j-th data, representing the transient part of the signal.

[0113] Calculate the mean of the upper envelope and the lower envelope to obtain the target mean:

[0114]

[0115] Among them, represents the target mean of the upper and lower envelopes, Env u (t) and Env l (t) respectively represent the upper envelope and the lower envelope, and t1 and t2 are the time intervals for calculating the mean.

[0116] Perform the first decomposition on the fused multi-source data based on the target mean to obtain multiple first decomposition data. For each, perform the first decomposition on the fused multi-source data based on the target mean: subtract the target mean from the original signal to obtain a new signal, which is the residual of the original signal after removing the local mean. When the residual still contains local maxima and minima, repeat removing the local mean, continue with envelope construction and target mean calculation until the residual no longer contains significant local extrema.

[0117] Embed white noise into the obtained first decomposition data to get second decomposition data, and perform averaging on all the second decomposition data to obtain multi-scale decomposition data.

[0118] Perform sparse coding on the multi-scale decomposition data to obtain coded decomposition data.

[0119] Perform sparse representation processing on the coded decomposition data to obtain sparse representation data.

[0120] S3: Perform dynamic time warping processing on the sparse representation data to obtain time series feature data.

[0121] Furthermore, segment the sparse representation data into different data points, each data point representing the device state at a moment, perform data point matching, and analyze the distance between adjacent data points for each data point to obtain the distance data f1(X a ,X b ) for each data point:

[0122]

[0123] Calculate the cumulative distance matrix F for each data point based on the distance data corresponding to each data point ab :

[0124]

[0125] Generate the backtracking path L for each data point through the cumulative distance matrix of each data point a :

[0126]

[0127] For example, starting from the lower right corner (n, m) of the cumulative distance matrix D(n, m), gradually trace back to the upper left corner (1, 1) D(1, 1). At each step, select the path with the minimum cumulative distance until reaching the starting point. In this way, the trace-back path for each data point can be generated. Based on the trace-back path of each data point, perform temporal feature extraction on the sparse representation data to obtain temporal feature data. The purpose of temporal feature extraction is to extract key features from the sparse representation data that reflect the operating state of the device, helping to identify the operating mode and potential faults of the device. Specifically, by analyzing the trace-back path, the best alignment of data points on the time axis can be found, thereby extracting the main features of the time series. For example, the time offset and cumulative distance change rate of each data point on the trace-back path can be calculated, and these features can reflect the dynamic changes of the device at different time points. In this way, a complete set of temporal feature data can be obtained, which can effectively describe the operating state and change trend of the GIS device.

[0128] Based on the trace-back path of each of the said data points, perform temporal feature extraction h(X, t) on the said sparse representation data to obtain the said temporal feature data Y a :

[0129]

[0130] where X a and X b are different data points after segmentation, representing the state of the GIS device at a specific moment. F ab represents the cumulative distance from data point X a to data point X b , X k is the intermediate data point during the accumulation process, L a is the trace-back path of the a-th data point, representing the optimal path from the current data point back to the starting point, g(a) represents the importance of the b-th data point, w is the total number of data points, t a and t a+1 respectively represent the time points corresponding to data points X a and X a+1 .

[0131] Through these temporal feature data, abnormal vibration patterns that may exist during the operation of the device can be identified, thereby timely discovering and locating potential faults of the device.

[0132] S4: Perform fractal dimension feature extraction on the said temporal feature data to obtain fractal feature data.

[0133] Furthermore, for the said temporal feature data Y aCalibrate the data range to obtain the target data range, and segment the time series feature data based on the target data range to obtain different time series feature data subsequences B with the same time interval T: B1{Y a1 , Y a2 , …Y ai}, B2{Y a(i+1) , Y a(i+2) , …Y a(i+j)}, …B n {Y a(i+j+1) , Y a(i+j+2) , …Y an}, and assign measurement values to each subsequence respectively. Among them, the measurement value is the energy fluctuation intensity, which is obtained by calculating the energy of the subsequence E = {Y a1 2 + Y a2 2 + …Y ai 2}, {Y a(i+1) 2 + Y a(i+2) 2 + … + Y a(i+j) 2}, …{Y a(i+j+1) 2 + Y a(i+j+2) 2 + … + Y an 2}.

[0134] Calculate the fractal dimension of each subsequence through the box dimension and multifractal spectrum method based on the measurement value of each subsequence to obtain the fractal dimension G(B) of each subsequence:

[0135]

[0136] Among them, A is the length of the subsequence, N is the number of sample points of the subsequence, and α is the weight for adjusting the measurement value.

[0137] Construct the multifractal spectrum f(γ) according to the fractal dimension of each subsequence:

[0138]

[0139] Among them, p(B[n]) is the normalized probability distribution of the subsequence B, δ is the impulse of the subsequence sample point, and γ is the weight index of the fractal spectrum.

[0140] Extract features from the multifractal spectrum to obtain the fractal feature data H:

[0141]

[0142] Among them, is the peak value of the multifractal spectrum, is the average magnitude of the signal scale change, is the degree of dispersion of the signal scale change, and <γ> is the expected value of γ.

[0143] By analyzing the multifractal spectrum, characteristic data reflecting the signal complexity and multi-scale characteristics can be extracted. For example, the peak value, spectral width, and other statistical characteristics of the multifractal spectrum can be calculated, and these characteristics can reflect the changes of the signal under different dimensions and scales. In this way, a complete set of fractal characteristic data can be obtained, which not only describes the overall complexity of the signal but also reflects the change characteristics of the signal at different time scales.

[0144] Suppose the temperature data of a GIS device is being analyzed. After dynamic time warping processing of this data, time series characteristic data is obtained. First, through data range calibration, the maximum and minimum values of the temperature data are determined. For example, the temperature range is between -10°C and 60°C. Next, the temperature data is segmented by month, and the data of each month is used as a subsequence.

[0145] Then, the energy fluctuation intensity is calculated for each subsequence. Suppose the data of a certain month is {30, 32, 35, 28, 31,...}, then its energy fluctuation intensity is: E = 30 2 + 32 2 + 35 2 + 28 2 + 31 2 +... Based on the energy fluctuation intensity of each subsequence, the fractal dimension of each subsequence is calculated. Suppose the calculated result of the box dimension is 1.2. Then, the multifractal spectrum of each subsequence is further calculated. Suppose the peak value of the multifractal spectrum of a certain month is 2.5 and the spectral width is 0.8. Through these calculations, the fractal characteristic data of each month is obtained, including the fractal dimension, the peak value and spectral width of the multifractal spectrum, etc.

[0146] S5: Perform spectral clustering processing on the fractal characteristic data to obtain clustering characteristic data.

[0147] Furthermore, calculate the feature similarity similarity(H i , H j ) of the fractal characteristic data H:

[0148] similarity(H i , H j ) = exp(-β||H i - H j )

[0149] Obtain a similarity dataset, where β is the bandwidth parameter of the Gaussian kernel, controlling the smoothness of the function.

[0150] Construct a Laplacian matrix based on the similarity dataset where C is the degree of each vertex in the Laplacian matrix, and L norm is the unnormalized Laplacian matrix, and perform normalization processing on the Laplacian matrix to obtain a normalized Laplacian matrix and perform eigenvalue decomposition to obtain multiple decomposed eigenvalues where V is the eigenvector, is the corresponding eigenvalue.

[0151] Perform spectral clustering on the decomposed eigenvalues to obtain clustering feature data K:

[0152]

[0153]

[0154] where, is the eigenvector matrix composed of the eigenvectors corresponding to the first m smallest eigenvalues, V m is the m-th eigenvector, S i is the i-th cluster.

[0155] S6: Input the clustering feature data into the local outlier factor analysis algorithm for outlier extraction to obtain outliers, and determine the device defect location data according to the outliers.

[0156] Furthermore, input the clustering feature data K into the local outlier factor analysis algorithm for local density analysis to obtain local density data ρ(K):

[0157]

[0158] where Ω is the spatial region of the data points, ‖K a -K b ‖ is the Euclidean distance between the point K a and K b and ε is the width parameter of the kernel function, f(K b ) is the clustering feature of the point K b .

[0159] Perform neighborhood density data matching on the local density data to obtain neighborhood density data ω(K):

[0160]

[0161] where P is the number of data points in the neighborhood, ψ is the parameter controlling the curvature of the function, r(K, K a)Used to compare the relationship between two data points.

[0162] Calculate the reachable distance η(K) of the local density data through the neighborhood density data:

[0163]

[0164] Where τ and ξ are parameters that control the shape and scale of the function.

[0165] Extract outliers from the clustering feature data through the reachable distance of the data points to obtain outliers, and determine the device defect location data according to the outliers: when the local density data ρ(K) of a data point is higher than 2.5, it means that the local density of the current data point is lower than the neighborhood density.

[0166] It should be noted that by performing the above steps, through a series of steps such as multi-source data fusion, multi-scale decomposition, sparse representation, dynamic time warping, fractal dimension feature extraction, spectral clustering, and local outlier factor analysis, the precise positioning of GIS equipment defects is achieved, which has significant beneficial effects. First, by collecting the original operation data of GIS equipment and performing multi-source data fusion, the comprehensiveness and diversity of the data are ensured, providing a reliable data basis for subsequent analysis and processing. Multi-source data fusion can effectively integrate data from different sensors, make up for the deficiencies of single-sensor data, and improve the richness and accuracy of the data. Secondly, in the multi-scale decomposition process, the empirical mode decomposition method is adopted, which can adaptively decompose non-linear and non-stationary signals and extract signal features at different time scales, ensuring the accuracy and meticulousness of the decomposition results. At the same time, through sparse representation processing, high-dimensional data is converted into low-dimensional sparse representation, effectively reducing the data dimension and improving the computational efficiency of subsequent processing and the robustness of data processing. The dynamic time warping processing step calculates the similarity of signals in different time periods in the sparse representation data, finds the optimal alignment path, and extracts time series feature data. This step can effectively align signals with time offsets and extract key features reflecting the dynamic changes of the signals, enhancing the accuracy and consistency of time series analysis. In the fractal dimension feature extraction process, by combining the box dimension and multifractal spectrum methods, the fractal dimension features of the time series feature data are calculated to obtain more detailed and diverse fractal feature data. The fractal dimension can reveal the internal complexity and multi-scale characteristics of the data, providing rich feature information for subsequent clustering analysis. The spectral clustering processing step maps high-dimensional data to a low-dimensional space through feature similarity calculation, construction and normalization of the Laplacian matrix, and eigenvalue decomposition method for clustering analysis to obtain clustering feature data. This method can effectively identify different patterns and structures in the data and improve the accuracy and stability of the clustering results. Finally, through local outlier factor (LOF) analysis, local density analysis and outlier detection are performed on the clustering feature data to identify outliers with significantly lower local density than the neighborhood, and finally determine the defect location data of the equipment. The LOF algorithm can effectively detect local outliers within the cluster, further improving the accuracy and reliability of defect location.

[0167] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not restrictive. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

[0168] Embodiment 2

[0169] An embodiment of the present invention provides a method for locating equipment defects of GIS equipment. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.

[0170] In this experiment, we selected 10 GIS devices, and they all showed varying degrees of performance degradation under normal operating conditions. These devices were randomly divided into two groups, one group using the traditional defect location method, and the other group using the GIS equipment defect location method proposed by the present invention.

[0171] Collect operation data of each group of devices for one month, including parameters such as temperature, humidity, voltage, and current; apply a weighted moving average function to process the original data to detect and eliminate outliers; fuse the processed data to form a multi-source data set; perform multi-scale decomposition and sparse representation processing on the fused data; extract time series features through dynamic time warping processing; calculate the fractal dimension using the box dimension and multifractal spectrum method; perform spectral clustering processing on the fractal feature data; apply the local outlier factor analysis algorithm to detect abnormal points and locate defects; show the comparison of the performance of the two groups of devices in defect location through Table 1:

[0172] Table 1

[0173]

[0174] It can be seen from the data table that the devices using the method of the present invention have significantly lower defect location time than the traditional method. For example, Device No. 1 using the traditional method takes 24 hours to locate the defect, while using the method of the present invention only takes 12 hours; the location accuracy rate of most devices using the method of the present invention exceeds 90%, while the accuracy rate of the traditional method is generally between 70% and 85%; in terms of the defect false alarm rate, the method of the present invention is also significantly lower than the traditional method. This shows that the method of the present invention not only improves the efficiency and accuracy of location, but also reduces the possibility of false alarms.

[0175] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not restrictive. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

[0176] Embodiment 3

[0177] The third embodiment of the present invention is different from the first two embodiments in that:

[0178] If the above-described functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0179] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in combination with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in combination with an instruction execution system, apparatus, or device.

[0180] More specific examples (non-exhaustive list) of computer-readable media include the following: electrical connection parts with one or more wirings (electronic devices), portable computer disk cartridges (magnetic devices), random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memories), fiber optic devices, and portable compact disc read-only memories (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or, if necessary, other appropriate processing, and then storing it in a computer memory.

[0181] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0182] Embodiment 4

[0183] Referring to Figure 2 , which is the fourth embodiment of the present invention. This embodiment provides a device defect location system for a GIS device, including an acquisition module, a decomposition module, a regularization module, an extraction module, a clustering module, and a location module.

[0184] The acquisition module is used to acquire the original operation data of a preset GIS device and convert the original operation data into fused multi-source data.

[0185] The decomposition module is used to perform multi-scale decomposition on the fused multi-source data to obtain multi-scale decomposition data, and perform sparse representation processing on the multi-scale decomposition data to obtain sparse representation data.

[0186] The regularization module is used to perform dynamic time warping processing on the sparse representation data to obtain time series feature data.

[0187] The extraction module is used to extract fractal dimension features from the time series feature data to obtain fractal feature data.

[0188] The clustering module is used to perform spectral clustering processing on the fractal feature data to obtain clustering feature data.

[0189] The location module is used to input the clustering feature data into a local outlier factor analysis algorithm for outlier extraction to obtain outliers, and determine device defect location data according to the outliers.

[0190] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A method for locating device defects of a GIS device, characterized in that: including collecting the original operation data of a preset GIS device and converting the original operation data into fused multi-source data performing multi-scale decomposition on the fused multi-source data to obtain multi-scale decomposition data, and performing sparse representation processing on the multi-scale decomposition data to obtain sparse representation data performing dynamic time warping processing on the sparse representation data to obtain time series feature data extracting fractal dimension features from the time series feature data to obtain fractal feature data performing spectral clustering processing on the fractal feature data to obtain clustering feature data inputting the clustering feature data into a local outlier factor analysis algorithm to extract outliers, and performing equipment defect detection based on the outliers inputting the clustering feature data K into a local outlier factor analysis algorithm to perform local density analysis, and obtaining local density data ρ(K): where Ω is the spatial region of data points, ‖K a -K b ‖ is the Euclidean distance between points K a and K b , ε is the width parameter of the kernel function, f(K b ) is the clustering feature of point K b ; performing neighborhood density data matching on the local density data to obtain neighborhood density data ω(K): where P is the number of data points in the neighborhood, ψ is the parameter that controls the curvature of the function, and r(K, K a ) is used to compare the relationship between two data points; calculating the reachable distance of data points η(K) through the neighborhood density data for the local density data where τ and ξ are parameters that control the shape and scale of the function extracting outliers from the clustering feature data through the reachable distance of data points, and performing equipment defect detection based on the outliers: when the local density data ρ(K) of a data point is higher than 2.5, it indicates that the local density of the current data point is lower than the neighborhood density 2. The device defect location method of a GIS device according to claim 1, characterized in that: The fused multi-source data includes collecting the operation data of sensors arranged by a pre-set GIS device during operation to obtain the original operation data D = {d1, d2, …, d t}, where d t represents the data at the t-th time point; the original operation data includes equipment status data, environmental data, and maintenance history data the equipment status data includes temperature, humidity, voltage, current, vibration, sound, insulating gas concentration, and circuit breaker operation times the environmental data includes external temperature, precipitation data, and air quality the maintenance history data includes maintenance records, fault history, and equipment structure parameters detecting outliers Detect(t) for the original operation data, and setting a weighted moving average function WMA(t): where W is the time window, T is the time to obtain data, and σ WMA is the standard deviation of the Gaussian kernel, which controls the speed of weight decay; monitoring abnormal data I(t) by accumulating abnormal scores within a T time period where S(t) is an abnormal score data function. When I(t) > θ and θ = 4, let Detect(t) be 1. When I(t) ≤ θ, let Detect(t) be 0. The final detection model is Anomaly(t)=Detect(t)·d t When Anomaly(t) = 0, it indicates that no anomaly is detected at time point t. When Anomaly(t) = d t it indicates that an anomaly is detected at time point t, and d t is the value of the abnormal data, obtaining the abnormal value data; classifying the outlier data according to features and attributes to obtain data dimensions with different attributes, and performing data fusion on the original operation data based on the data dimensions to obtain fused multi-source data 3. The device defect location method of a GIS device according to claim 2, characterized in that: the sparse representation data includes inputting the fused multi-source data into a preset empirical mode decomposition algorithm to construct a signal envelope Env(t), where the signal envelope includes an upper envelope and a lower envelope Among them, EMD is the empirical mode decomposition function for decomposing multi-source data. x and y respectively represent the components of the signal in the horizontal and vertical directions, and μ i and σ i are respectively the mean and variance of the Gaussian distribution of the i-th data, representing the continuous part of the signal; λ j and ν j are respectively the rate parameter and location parameter of the exponential distribution of the j-th data, representing the transient part of the signal; calculating the mean of the upper envelope and the lower envelope to obtain a target mean Among them, represents the target mean of the upper and lower envelopes, Env u (t) and Env l (t) represent the upper envelope and the lower envelope respectively, and t1 and t2 are the time intervals for calculating the mean; Perform the first decomposition on the fused multi-source data based on the target mean to obtain multiple first decomposition data. For each, perform the first decomposition on the fused multi-source data based on the target mean: Subtract the target mean from the original signal to obtain a new signal, which is the residual after removing the local mean from the original signal. When the residual still contains local maxima and minima, repeat removing the local mean, continue with envelope construction and target mean calculation until the residual no longer contains significant local extrema; Embed white noise into the obtained first decomposition data to obtain second decomposition data, and perform averaging processing on all the second decomposition data to obtain multi-scale decomposition data; Perform sparse coding on the multi-scale decomposition data to obtain coded decomposition data; Perform sparse representation processing on the coded decomposition data to obtain sparse representation data.

4. The device defect location method of a GIS device according to claim 3, characterized in that: The dynamic time warping process includes segmenting the sparse representation data into different data points, where each data point represents the device state at a moment, performing data point matching, and respectively analyzing the distances between adjacent data points for each data point to obtain the distance data f1(X a ,X b ): Calculate the cumulative distance matrix F for each data point based on the distance data corresponding to each of the said data points ab : Generate the backtracking path L for each data point through the cumulative distance matrix of each data point a : Based on the backtracking path of each of the said data points, perform temporal feature extraction h(X, t) on the sparse representation data to obtain the temporal feature data Y a : Among them, X a and X b are different data points to be segmented, representing the state of the GIS device at a specific moment; F ab represents the cumulative distance from the data point X a to the data point X b , X k is the intermediate data point during the accumulation process, L a is the backtracking path of the a-th data point, representing the optimal path from the current data point back to the starting point, g(a) represents the importance of the b-th data point, w is the total number of data points, t a and t a+1 respectively represent the time points corresponding to the data points X a and X a+1 .

5. The device defect location method of a GIS device according to claim 4, characterized in that: The fractal dimension feature extraction includes the time series feature data Y a Performing data range calibration to obtain a target data range, and based on the target data range, segmenting the time series feature data to obtain different time series feature data subsequences B with the same time interval T: B1Y a1 ,Y a2 ,…Y ai 、B2Y ai+1 ,Y ai+2 ,…Y ai+j …B n Y ai+j+1 ,Y ai+j+2 ,…Y an , and respectively assigning measurement values to each subsequence, where the measurement value is the energy fluctuation intensity, obtained by calculating the energy of the subsequence E = Y a1 2 +Y a2 2 +…Y ai 2 、Y ai+1 2 +Y ai+2 2 +…+Y ai+j 2 …Y ai+j+1 2 +Y ai+j+2 2 +…+Y an 2 ; Calculate the fractal dimension of each subsequence through the methods of box dimension and multifractal spectrum based on the measurement values of each subsequence to obtain the fractal dimension G(B) of each subsequence: where A is the length of the subsequence, N is the number of sample points of the subsequence, and α is the weight for adjusting the measurement value; Construct the multifractal spectrum f(γ) according to the fractal dimension of each subsequence: where p(B[n]) is the normalized probability distribution of the subsequence B, δ is the impulse of the subsequence sample points, and γ is the weight index of the fractal spectrum; Perform feature extraction on the multifractal spectrum to obtain the fractal feature data H; Among them, is the peak value of the multifractal spectrum, is the average size of the signal scale change, is the degree of dispersion of the signal scale change, and <γ> is the expected value of γ.

6. The device defect location method of a GIS device according to claim 5, characterized in that: The spectral clustering process includes calculating the feature similarity similarity(H i ,H j ) for the fractal feature data H: similarity(H i ,H j )=exp(-β||H i -H j ) Obtain the similarity data set, where β is the bandwidth parameter of the Gaussian kernel, controlling the smoothness of the function; Construct a Laplacian matrix based on the similarity dataset where C is the degree of each vertex in the Laplacian matrix, and L norm is the unnormalized Laplacian matrix, and the Laplacian matrix is normalized to obtain a normalized Laplacian matrix and eigenvalue decomposition is performed on the normalized Laplacian matrix to obtain multiple decomposed eigenvalues where V is the eigenvector, is the corresponding eigenvalue; Perform spectral clustering processing on the decomposed eigenvalue to obtain clustering feature data K; U Lnorm = [V1, V2,..., V m ​ Among them, is an eigenvector matrix composed of eigenvectors corresponding to the first m smallest eigenvalues, V m is the m-th eigenvector, S i is the i-th cluster.

7. A system for a device defect location method using a GIS device as described in any one of claims 1 to 6, characterized in that: Including an acquisition module, a decomposition module, a regularization module, an extraction module, a clustering module, and a positioning module; The acquisition module is used to acquire the original operation data of the preset GIS device and convert the original operation data into fused multi-source data; The decomposition module is used to perform multi-scale decomposition on the fused multi-source data to obtain multi-scale decomposition data, and perform sparse representation processing on the multi-scale decomposition data to obtain sparse representation data; The regularization module is used to perform dynamic time warping processing on the sparse representation data to obtain time series feature data; The extraction module is used to perform fractal dimension feature extraction on the time series feature data to obtain fractal feature data; The clustering module is used to perform spectral clustering processing on the fractal feature data to obtain clustering feature data; The positioning module is used to input the clustering feature data into the local outlier factor analysis algorithm for outlier extraction to obtain outliers, and perform equipment defect detection based on the outliers.

8. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method according to any one of claims 1 to 6.

Images