GIS equipment detection system and method based on partial discharge-vibration signal fusion

Abstract

The invention provides a GIS equipment detection system and method based on partial discharge-vibration signal fusion, and relates to the technical field of GIS equipment detection. The method has the flexibility of adapting to different data conditions, and vibration signals and partial discharge signals at a contact and a basin-type insulator are respectively obtained through blind source separation of collected multi-source signals for GIS equipment fault detection; the respective sequential relationship of the vibration signal and the partial discharge signal is considered, the multi-scale fuzzy entropy of the vibration signal and the partial discharge signal is calculated, the richness represented by the vibration characteristic and the partial discharge characteristic is further enhanced, the problems of poor reliability and large detection error of a fault detection method based on a single technology are effectively solved, and the fault detection efficiency is improved. The problems that the characteristics of an existing vibration diagnosis system and the characteristics of an existing partial discharge diagnosis system are prone to omission and low in accuracy are solved. Meanwhile, a three-dimensional digital twinborn visual model of the GIS equipment is also established, and the operation of the GIS equipment can be monitored in real time in combination with fault information.

Description

Technical Field

[0001] The present invention relates to the technical field of GIS equipment detection, and particularly to a GIS equipment detection system and method based on the fusion of partial discharge - vibration signals. Background Art

[0002] As a gas-insulated switchgear, GIS equipment is widely used in high-voltage power systems due to its advantages such as compact structure, high reliability, flexible configuration, and small maintenance workload. Substation GIS equipment has various potential faults during long-term high-load operation. If these faults are not detected and processed in time, they may cause equipment damage and affect the stable operation of the power system.

[0003] The fault types of GIS equipment include partial discharge, mechanical faults, SF6 gas leakage, etc. Partial discharge will damage the insulation of GIS, cause conductor heating and abnormal noises, threatening the normal operation of the equipment. Mechanical faults will cause abnormal vibrations of GIS equipment, which may trigger mechanical faults of other primary equipment such as the refusal and misoperation of circuit breakers and disconnectors, affecting the normal operation of the equipment. At present, most of the detection methods for GIS equipment faults use a single type of sensor to detect and diagnose a single physical quantity, such as ultra-high frequency, ultrasonic detection, etc., and less use the method of comprehensive multi-parameter detection to study the change process and law of the operating state of GIS equipment during operation. These methods have their own limitations, such as signal attenuation and small effective signal range during actual detection, and the signal data collected by a single sensor is extremely vulnerable to environmental noise. How to apply multiple types of sensors, and organically integrate the sensors and the corresponding data acquisition and signal transmission systems to perform efficient state analysis and fault diagnosis on the equipment will be an important research direction for GIS equipment fault detection technology. Although the existing GIS equipment fault diagnosis methods based on partial discharge and vibration detection have good effects in specific fields, they also have deficiencies such as complex signals, difficult fault location, limited detection range, and inability to monitor the operating state.

[0004] With the development of technologies such as intelligence and big data, GIS equipment fault diagnosis technology is also developing towards intelligence, remote control, integration, etc. Therefore, there is an urgent need for a GIS equipment fault detection system based on multi-technology fusion to improve the accuracy of GIS equipment fault detection while realizing real-time monitoring of the operating state of GIS equipment. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a GIS equipment detection system and method based on the fusion of partial discharge - vibration signals for detecting the fault types and locations of GIS equipment and real-time monitoring of the operating state of GIS equipment in view of the above-mentioned deficiencies of the prior art.

[0006] To solve the above technical problems, the technical solution adopted by the present invention is as follows: On the one hand, the present invention provides a GIS device detection system based on the fusion of partial discharge - vibration signals, including: a vibration signal acquisition module, a partial discharge signal acquisition module, a vibration signal feature analysis module, a partial discharge signal feature analysis module, and a fault diagnosis and monitoring module.

[0007] The vibration signal acquisition module is used to acquire the vibration signals of the GIS device. By connecting the contacts of multiple vibration sensors to different positions to be measured of the GIS device, the vibration signals of the GIS device are acquired, and the acquired vibration signals of the GIS device are pre - processed. Through the blind source separation algorithm, the independent vibration data collected at the contacts of each vibration sensor and at the pot - type insulators are obtained respectively, and are transmitted to the vibration signal feature analysis module;

[0008] The partial discharge signal acquisition module is used to acquire the partial discharge signals of the GIS device. By connecting the contacts of multiple partial discharge sensors to different positions to be measured of the GIS device, the partial discharge signals of the GIS device are acquired, and the acquired partial discharge signals of the GIS device are pre - processed. Through the blind source separation algorithm, the independent partial discharge data collected at the contacts of each partial discharge sensor and at the pot - type insulators are obtained respectively, and are transmitted to the partial discharge signal feature analysis module;

[0009] The vibration signal feature analysis module is used to calculate and extract the vibration features of the GIS device, and transmit the vibration feature data to the fault diagnosis and monitoring module;

[0010] The partial discharge signal feature analysis module is used to convert different types of partial discharge signals from analog signals to digital signals, calculate and extract the partial discharge features of the GIS device, and transmit the partial discharge feature data to the fault diagnosis and monitoring module;

[0011] The fault diagnosis and monitoring module is used to identify and locate the fault types and location information of the GIS device contained in the feature data obtained by the vibration signal feature analysis module and the partial discharge signal feature analysis module, store the identified fault information and issue a fault alarm, and establish a three - dimensional digital twin visualization model of the GIS device to realize the operation monitoring of the GIS device.

[0012] Preferably, the vibration signal acquisition module includes a multi-channel vibration sensor sequence composed of multiple vibration sensors, a vibration signal data preprocessing sub-module, and a first EMD-ICA blind source separation sub-module. The multi-channel vibration sensor sequence includes multiple vibration sensors installed at different positions to be measured on the GIS device. The contacts of the multiple vibration sensors are connected to different positions to be measured on the GIS device. The vibration signal data preprocessing sub-module is used to improve the quality of the vibration signal and ensure that the data between different vibration sensors has a consistent scale. The first EMD-ICA blind source separation sub-module is used to separate the vibration sensor contact vibration data and the independent vibration data of the pot insulator collected by each vibration sensor and transmit them to the vibration signal feature analysis module.

[0013] Preferably, the partial discharge signal acquisition module includes a multi-channel partial discharge sensor sequence composed of multiple partial discharge sensors, a partial discharge signal data preprocessing sub-module, and a second EMD-ICA blind source separation sub-module. The multi-channel partial discharge sensor sequence includes multiple partial discharge sensors installed at different positions to be measured on the GIS device. The contacts of the multiple partial discharge sensors are connected to different positions to be measured on the GIS device. The partial discharge signal data preprocessing sub-module is used to improve the quality of the partial discharge signal and ensure that the data between different partial discharge sensors has a consistent scale. The second EMD-ICA blind source separation sub-module is used to separate the partial discharge sensor contact partial discharge data and the independent partial discharge data of the pot insulator collected by each partial discharge sensor and transmit them to the partial discharge signal feature analysis module.

[0014] Preferably, the fault diagnosis and monitoring module includes a fault diagnosis sub-module, a data storage and alarm device, and a three-dimensional digital twin visual operation model construction sub-module. The fault diagnosis sub-module is used to identify and locate the fault type and location information of the GIS device contained in the feature data obtained by the vibration feature analysis module and the partial discharge feature analysis module. The data storage and alarm device is used to store the fault information and issue a fault alarm. The three-dimensional digital twin visual operation model construction sub-module is used to generate a visual model of the GIS device.

[0015] On the other hand, the present invention also provides a method for detecting GIS devices based on the fusion of partial discharge-vibration signals, including the following steps:

[0016] Step 1: The vibration signal acquisition module and the partial discharge signal acquisition module collect and separate the independent vibration signals and independent partial discharge signals at different positions to be measured on the GIS device;

[0017] Step 1.1: According to the requirements of GIS equipment detection tasks, arrange multiple vibration sensors in the multi-channel vibration signal sensor sequence and multiple partial discharge sensors in the multi-channel partial discharge sensor sequence at different positions to be measured on the GIS equipment, so as to comprehensively collect the vibration signals and partial discharge signals at different positions to be measured on the GIS equipment, including the vibration signals at the contacts of the vibration sensors, the vibration signals at the pot-type insulators, the partial discharge signals at the contacts of the partial discharge sensors, and the partial discharge signals at the pot-type insulators;

[0018] Step 1.2: Preprocess the collected vibration signals and partial discharge signals of the GIS equipment to improve the signal quality and ensure that the data between different samples has a consistent scale;

[0019] Step 1.3: Use the EMD-ICA blind source separation algorithm to separate and screen out the independent vibration signals at the contacts and pot-type insulators of each vibration sensor, and the independent partial discharge signals at the contacts and pot-type insulators of each partial discharge sensor, and extract the initial features of the vibration signals at the contacts and pot-type insulators of the GIS equipment vibration sensors, and the initial features of the partial discharge signals at the contacts and pot-type insulators of each partial discharge sensor;

[0020] Step 2: The vibration signal feature extraction module uses the MFE algorithm to calculate and extract the vibration features of the independent vibration signals at the contacts and pot-type insulators at different positions to be measured on the GIS equipment;

[0021] Step 2.1: Coarsely granulate the vibration signals after blind source separation, and divide the vibration signals into several segments using the time scale factor;

[0022] The vibration signal y i (t) after blind source separation is coarsely granulated, and the vibration signal is divided into several segments using the time scale factor τ. Let 1 ≤ j ≤ N, and divide y i into τ segments, each segment with a length of N / τ, as shown in the following formula:

[0023]

[0024] Step 2.2: Use the MFE algorithm to calculate the fuzzy entropy of each segment of the vibration signal, and select the component with the smallest entropy as the vibration feature of the GIS equipment;

[0025] Select different time scale factors τ to calculate the fuzzy entropy FuzzyEn of the characteristic signal. Given the embedding dimension c, the similarity tolerance r, and the fuzzy function gradient grad, calculate the fuzzy entropy of each segment of the coarsely granulated signal, as shown in the following formula:

[0026]

[0027] Select the component with the smallest entropy as the vibration feature p;

[0028] Step 3: The partial discharge signal feature extraction module uses the MFE algorithm to calculate and extract the partial discharge features of the independent partial discharge signals at the contacts and potting insulators at different positions to be measured of the GIS device;

[0029] Step 3.1: Convert different types of partial discharge signals from analog signals to digital signals through an analog-to-digital converter as partial discharge feature signals;

[0030] Step 3.2: Use the same method as in Step 2 for extracting the vibration features of the GIS device to extract the partial discharge feature q of the partial discharge feature signal;

[0031] Step 4: The fault diagnosis and monitoring module classifies the fault information contained in the GIS device according to the vibration features and partial discharge features, identifies the fault type and location of the GIS device, establishes a three-dimensional digital twin visualization model of the GIS device, and realizes visual monitoring of the operation of the GIS device in combination with the fault information:

[0032] Step 4.1: Based on the vibration features and partial discharge features obtained by the vibration signal feature analysis module and the partial discharge signal feature analysis module, use the sparse coding algorithm to achieve fault classification of the GIS device;

[0033] Construct a feature data matrix G using the vibration feature p and the partial discharge feature q, where each column represents a sample, each row represents a feature, and it contains m features and n samples in total; Use the K-SVD algorithm to generate an initial dictionary, and the number of atoms contained in the dictionary and the dimension of each atom in the dictionary are determined according to the characteristics and requirements of the data, and learn a dictionary matrix D ∈ R m ×κ , where R is the set of all possible values of each atom in the dictionary matrix D, and k is the number of basis vectors of the dictionary;

[0034] Find a sparse coefficient matrix B ∈ R k×n such that each sample x i can be represented as a linear combination of dictionary vectors, as shown in the following formula:

[0035] x i ≈ Db i

[0036] where b i is the sparse coding of the sample x i ;

[0037] Iteratively perform sparse coding and dictionary update until convergence or reaching the preset number of iterations, obtain the updated dictionary D′ and the sparse coefficient matrix B′, reconstruct the feature data matrix G, obtain the reconstructed feature data matrix G′, and evaluate the effect of the sparse coding algorithm through the reconstruction error to verify the effectiveness of the sparse coding algorithm;

[0038] Use the sparse coding algorithm to reconstruct the feature data matrix G to obtain the reconstructed feature data matrix G'. Classify the GIS equipment faults into different types, including contact electrical contact faults, contact partial discharge faults, pot insulator mechanical faults, pot insulator breakdown faults, and pot insulator flashover faults.

[0039] Step 4.2: Combine the vibration characteristics and partial discharge characteristics to comprehensively judge the GIS equipment fault information, including fault feature matching and fault location.

[0040] Input the vibration characteristics and partial discharge characteristic data to be detected into the SVM model for fault feature matching to obtain the fault feature matching data, and use the sparse coding algorithm adopted in Step 4.1 to output the category labels of the corresponding fault types.

[0041] Input the vibration characteristics and partial discharge characteristic data to be detected into the EKF algorithm for fault location. By establishing a fault propagation model of the GIS equipment, describe the propagation characteristics of vibration signals and partial discharge signals and the changes during fault occurrence, use the EKF algorithm to process the vibration characteristics and partial discharge characteristic data, and obtain the estimated value of the fault location by iteratively updating the estimated value of the state variable.

[0042] Based on the fault feature matching data and the fault location data, judge the fault result of the GIS equipment, store the diagnosed fault information in the data storage and alarm device, and issue a fault alarm.

[0043] Step 4.3: Use the inverse process interactive modeling method to reconstruct and output the texture and geometric information of the GIS equipment, establish a three-dimensional digital twin visualization operation model of the GIS equipment, and realize visual monitoring of the operation of the GIS equipment in combination with the fault information.

[0044] The beneficial effects of adopting the above technical solutions are as follows: The GIS equipment detection system and method based on the fusion of partial discharge - vibration signals provided by the present invention have the flexibility to adapt to different data situations, consider the time sequence relationship between vibration data and partial discharge data, enhance the richness of feature representation, effectively solve the problems of poor reliability and large detection errors of the fault detection method based on a single technology, and at the same time overcome the problems of easy omission and low accuracy of the features of the existing vibration diagnosis system and partial discharge diagnosis system. At the same time, the present invention also establishes a three-dimensional digital twin visualization model of the GIS equipment, which can realize real-time monitoring of the operation of the GIS equipment in combination with the fault information. Description of the Drawings

[0045] Figure 1 Block diagram of the GIS equipment detection system based on the fusion of partial discharge - vibration signals provided by the embodiments of the present application;

[0046] Figure 2 This is the operation monitoring diagram of the GIS device simulated by the three-dimensional digital twin visualization model provided by the embodiments of the present invention. Specific Embodiments

[0047] The following will further describe in detail the specific embodiments of the present invention in conjunction with the accompanying drawings and embodiments. The following embodiments are used to illustrate the present invention, but are not used to limit the scope of the present invention.

[0048] In this embodiment, a GIS device detection system based on the fusion of partial discharge-vibration signals, as Figure 1 shown, includes a vibration signal acquisition module, a partial discharge signal acquisition module, a vibration signal feature analysis module, a partial discharge signal feature analysis module, and a fault diagnosis and monitoring module.

[0049] The vibration signal acquisition module is used to acquire the vibration signals of the GIS device. By connecting the contacts of multiple vibration sensors to different positions to be measured of the GIS device, the vibration signals of the GIS device are acquired, and the acquired vibration signals of the GIS device are preprocessed. Independent vibration data acquired at the contacts of each vibration sensor and at the pot-type insulators are obtained through a blind source separation algorithm and transmitted to the vibration signal feature analysis module;

[0050] The partial discharge signal acquisition module is used to acquire the partial discharge signals of the GIS device. By connecting the contacts of multiple partial discharge sensors to different positions to be measured of the GIS device, the partial discharge signals of the GIS device are acquired, and the acquired partial discharge signals of the GIS device are preprocessed. Independent partial discharge data acquired at the contacts of each partial discharge sensor and at the pot-type insulators are obtained through a blind source separation algorithm and transmitted to the partial discharge signal feature analysis module;

[0051] The vibration signal feature analysis module is used to calculate and extract the vibration features of the GIS device and transmit the vibration feature data to the fault diagnosis and monitoring module;

[0052] The partial discharge signal feature analysis module is used to convert different types of partial discharge signals from analog signals to digital signals, calculate and extract the partial discharge features of the GIS device, and transmit the partial discharge feature data to the fault diagnosis and monitoring module;

[0053] The fault diagnosis and monitoring module is used to identify and locate the fault types and location information of the GIS device contained in the feature data obtained by the vibration signal feature analysis module and the partial discharge signal feature analysis module, store the identified fault information and issue a fault alarm, and establish a three-dimensional digital twin visualization model of the GIS device to realize the operation monitoring of the GIS device;

[0054] The vibration signal acquisition module includes a multi-channel vibration sensor sequence composed of multiple vibration sensors, a vibration signal data preprocessing sub-module, and a first EMD-ICA blind source separation sub-module. The multi-channel vibration sensor sequence contains multiple vibration sensors installed at different positions to be measured on the GIS device. The contacts of the multiple vibration sensors are connected to different positions to be measured on the GIS device. The vibration signal data preprocessing sub-module is used to improve the quality of vibration signals and ensure that the data between different vibration sensors has a consistent scale. The first EMD-ICA blind source separation sub-module is used to separate the vibration data of the vibration sensor contacts and the independent vibration data of the pot insulator collected by each vibration sensor, and transmit them to the vibration signal feature analysis module;

[0055] The partial discharge signal acquisition module includes a multi-channel partial discharge sensor sequence composed of multiple partial discharge sensors, a partial discharge signal data preprocessing sub-module, and a second EMD-ICA blind source separation sub-module. The multi-channel partial discharge sensor sequence contains multiple partial discharge sensors installed at different positions to be measured on the GIS device. The contacts of the multiple partial discharge sensors are connected to different positions to be measured on the GIS device. The partial discharge signal data preprocessing sub-module is used to improve the quality of partial discharge signals and ensure that the data between different partial discharge sensors has a consistent scale. The second EMD-ICA blind source separation sub-module is used to separate the partial discharge data of the partial discharge sensor contacts and the independent partial discharge data of the pot insulator collected by each partial discharge sensor, and transmit them to the partial discharge signal feature analysis module;

[0056] Based on the first EMD-ICA blind source separation sub-module, the vibration signal acquisition module uses the EMD-ICA algorithm to perform blind source separation on the collected vibration signals of the GIS device, adaptively decomposes the collected composite vibration signals of the GIS device into a group of intrinsic mode functions (IMFs) with different center frequencies, respectively obtains the vibration data at the contacts of each vibration sensor and at the pot insulator, and transmits them to the vibration signal feature analysis module;

[0057] Based on the second EMD-ICA blind source separation sub-module, the partial discharge signal acquisition module uses the EMD-ICA algorithm to perform blind source separation on the obtained intrinsic mode functions, adaptively decomposes the collected composite partial discharge signals of the GIS device into a group of intrinsic mode functions with different center frequencies, respectively obtains the partial discharge data at the contacts of each partial discharge sensor and at the pot insulator, and transmits them to the partial discharge signal feature analysis module.

[0058] The fault diagnosis and monitoring module includes a fault diagnosis sub-module, a data storage and alarm device, and a three-dimensional digital twin visualization operation model construction sub-module. The fault diagnosis sub-module is used to identify and locate the fault types and location information of GIS equipment contained in the characteristic data obtained by the vibration characteristic analysis module and the partial discharge characteristic analysis module. The data storage and alarm device is used to store fault information and issue a fault alarm. The three-dimensional digital twin visualization operation model construction sub-module is used to generate a visualization model of the GIS equipment.

[0059] In this embodiment, a detection method for GIS equipment based on the fusion of partial discharge and vibration signals includes the following steps:

[0060] Step 1: The vibration signal acquisition module and the partial discharge signal acquisition module collect and separate independent vibration signals and independent partial discharge signals at different positions to be measured of the GIS equipment:

[0061] Step 1.1: According to the requirements of the GIS equipment detection task, a plurality of vibration sensors in the multi-channel vibration signal sensor sequence and a plurality of partial discharge sensors in the multi-channel partial discharge sensor sequence are arranged at different positions to be measured of the GIS equipment to comprehensively collect the vibration signals and partial discharge signals at different positions to be measured of the GIS equipment, including the vibration signals at the contacts of the vibration sensors, the vibration signals at the pot-type insulators, the partial discharge signals at the contacts of the partial discharge sensors, and the partial discharge signals at the pot-type insulators;

[0062] For GIS equipment that cannot be directly accessed or for which insufficient high-quality data can be collected, existing partial discharge and vibration data sets at the contacts and pot-type insulators of GIS equipment can be used for experiments and analysis.

[0063] Step 1.2: Preprocess the collected vibration signals and partial discharge signals of the GIS equipment to improve the signal quality and ensure that the data between different samples has a consistent scale;

[0064] By improving the signal quality and ensuring that the data between different samples has a consistent scale, there is a large amount of noise or signal distortion during the operation of the GIS equipment, and effective vibration data cannot be clearly collected. The voice enhancement technology can improve the quality of the vibration signal by means of noise reduction, enhancing relevant features of the vibration, etc. In this embodiment, the collected signals are first preprocessed using a noise removal filter and downsampling method; then, the voice enhancement technology is used to improve the quality of the collected vibration signals to obtain the original signal f(t) after preprocessing;

[0065] Step 1.3: Use the EMD-ICA blind source separation algorithm to separate and screen out the independent vibration signals at the contacts and pot-type insulators of each vibration sensor, and the independent partial discharge signals at the contacts and pot-type insulators of each partial discharge sensor, and respectively extract the initial features of the vibration signals at the contacts and pot-type insulators of the GIS device vibration sensors, and the initial features of the partial discharge signals at the contacts and pot-type insulators of each partial discharge sensor;

[0066] The EMD (Empirical Mode Decomposition) algorithm is a time-frequency domain signal processing method that decomposes signals based on the time-scale characteristics of the data itself. Use the EMD algorithm to decompose the original signal f(t) preprocessed in Step 1.2, and obtain several IMF components, including the following steps:

[0067] (1) Find all the maximum and minimum points in the original signal f(t) sequence, use the cubic spline interpolation method to outline the extreme points, and obtain the upper envelope and the lower envelope respectively, and then take their average value to obtain the average value function;

[0068] (2) Calculate the difference between the original signal f(t) and the average value function, and denote it as the i-th intermediate function h i (t);

[0069] (3) Determine whether the intermediate function h i (t) obtained in step (2) satisfies the two conditions of the IMF component: within the entire time range of the intermediate function, the number of local extreme points and zero-crossing points must be equal or differ by at most one; at any time point, the average of the envelope of the local maximum (upper envelope) and the envelope of the local minimum (lower envelope) must be zero;

[0070] If the two conditions of the IMF component are satisfied, the i-th intermediate function h i (t) is the i-th intrinsic mode function x i (t) of the original signal f(t); if not, take the i-th intermediate function h i (t) as the "original signal", and repeatedly execute steps (1) and (2) until the two IMF conditions are satisfied;

[0071] (4) Separate the i-th intrinsic mode function x i (t) from the original signal f(t) to obtain the i-th difference signal Res i (t);

[0072] (5) Take the i-th residual signal Res i (t) as the new "original signal" and repeat steps (1) to (4) until the n-th residual signal Res nStop when (t) is a monotonic function or a constant, and finally obtain the decomposed signal f′(t), as shown in the following formula:

[0073] f′(t) = ∑x n (t) + Res n (t)

[0074] where x n (t) is the nth IMF component;

[0075] The ICA (Independent Component Analysis) algorithm is a blind source separation technology that decouples complex mixed signals and recovers source signals from observed signals. The IMF components obtained by using the EMD algorithm can simplify the structure of the original signal. However, using the EMD algorithm alone is often insufficient to effectively separate the noise and useful information in the signal, and the adaptively obtained IMF components may also be redundant.

[0076] In this embodiment, the ICA algorithm is used to further extract independent components from the IMF components obtained by decomposing with the EMD algorithm, and it can more effectively filter out noise components. Denote the IMF components after decomposition by the EMD algorithm as the original observed signal X = (x1, x2,... x n ) T , and denote the source signal to be recovered as S = (s1, s2,... s n ) T , then the ICA model is shown in the following formula:

[0077] X = AS

[0078] where A is the mixing matrix;

[0079] Selecting an appropriate demixing matrix W can further extract independent IMF components to obtain the estimated signal Y = (y1, y2,... y n ) T , that is

[0080]

[0081] Evaluate the mutual independence between components by measuring the non-Gaussianity of the separation result, including the following steps:

[0082] 1) Center and whiten the original observed signal X:

[0083] Calculate the mean μ of the original observed signal X, as shown in the following formula:

[0084]

[0085] where N is the number of data points of the original observed signal;

[0086] Subtract the mean μ from the original observed signal X to obtain the centered signal X centered , as shown in the following formula:

[0087] X centered = X - μ

[0088] Calculate the covariance matrix Ω of the centered signal X centered , as shown in the following formula:

[0089]

[0090] Perform eigenvalue decomposition on the covariance matrix Ω, as shown in the following formula:

[0091] Ω = UΛU T

[0092] where U is the eigenvector matrix and Λ is the eigenvalue diagonal matrix;

[0093] Calculate the whitening matrix V, as shown in the following formula:

[0094]

[0095] Transform the centered signal X centered through the whitening matrix V to obtain the whitened signal X whitened , as shown in the following formula:

[0096] X whitened = VX centered

[0097] The covariance matrix of the whitened signal X whitened is the identity matrix;

[0098] 2) Set l as the number of channels of the signals to be separated, and initialize the iteration number k = 1;

[0099] 3) Randomly select a vector w k of the demixing matrix W as the initial vector;

[0100] 4) Use Newton's method to calculate w′ k , as shown in the following formula:

[0101] w′ k = E[Xg(w k T X)] - E[g′(w k T X)]w k and normalize it, as shown in the following formula:

[0102] w″k = w′ k / ||w′ k ||

[0103] where E[·] is the mathematical expectation;

[0104] The voice signal g(w k T X) is a super-Gaussian signal, as shown in the following formula:

[0105] g(w k T X) = w k T X exp(-(w k T X) 2 / 2)

[0106] where g′(w k T X) is the first derivative of the voice signal g(w k T X);

[0107] 5) Standardize and centralize w″ k ;

[0108] Calculate the mean μ k of each column of the vector w″ b , as shown in the following formula;

[0109]

[0110] where N′ is the number of rows of the vector w′ k ′, that is, the number of data points, a is the row index, and b is the column index;

[0111] Subtract the mean of each column from each element to obtain the centralized matrix W centered (a, b), as shown in the following formula:

[0112] W centered (a, b) = W(a, b) - μ b

[0113] Calculate the standard deviation σ k of each column of the vector w′ b , as shown in the following formula:

[0114]

[0115] where μ b is the mean of each column of the vector w′ k ′;

[0116] Divide each element of each column by the standard deviation of that column to obtain the standardized matrix W standardized (a,b):

[0117]

[0118] 6) Judge the convergence of the vector w′ k ′. If it converges, proceed to the next step; if not, return to step (4);

[0119] 7) Let k = k + 1. If k ≤ l, return to step (3); otherwise, the algorithm ends, obtaining the demixing matrix W, and then extracting the initial features of the vibration data and partial discharge data at the contact points of each sensor connected to different positions of the GIS device;

[0120] Step 2: The vibration signal feature extraction module uses the MFE algorithm to calculate and extract the vibration features of the independent vibration signals at different positions to be measured of the GIS device:

[0121] Step 2.1: Coarsely grain the vibration signal after blind source separation, and divide the vibration signal into several segments using the time scale factor;

[0122] The vibration signal y i (t) after coarse-grained blind source separation is divided into several segments using the time scale factor τ. Let 1 ≤ j ≤ N, and divide y i into τ segments, each segment having a length of N / τ, as shown in the following formula:

[0123]

[0124] Step 2.2: Use the MFE algorithm to calculate the fuzzy entropy of each segment of the vibration signal, and select the component with the minimum entropy as the vibration feature of the GIS device;

[0125] Select different time scale factors τ to calculate the fuzzy entropy FuzzyEn of the characteristic signal. Given the embedding dimension c, similarity tolerance r, and fuzzy function gradient grad, calculate the fuzzy entropy of each segment of the coarsely grained signal, as shown in the following formula:

[0126]

[0127] Select the component with the minimum entropy as the vibration feature p;

[0128] Step 3: The partial discharge signal feature extraction module uses the MFE algorithm to calculate and extract the partial discharge features of the independent partial discharge signals at different positions to be measured of the GIS device:

[0129] Step 3.1: Convert different types of partial discharge signals from analog signals to digital signals through an analog-to-digital converter as partial discharge characteristic signals;

[0130] Step 3.2: Use the same method as in Step 2 to extract the partial discharge feature q of the partial discharge feature signal;

[0131] Step 4: The fault diagnosis and monitoring module classifies the fault information contained in the GIS device according to the vibration feature and the partial discharge feature, identifies the fault type and location of the GIS device, establishes a three-dimensional digital twin visualization model of the GIS device, and realizes visual monitoring of the operation of the GIS device in combination with the fault information:

[0132] Step 4.1: Based on the vibration feature and the partial discharge feature obtained by the vibration signal feature analysis module and the partial discharge signal feature analysis module, adopt the sparse coding algorithm to realize the fault classification of the GIS device;

[0133] Construct a feature data matrix G using the vibration feature p and the partial discharge feature q, where each column represents a sample, each row represents a feature, and it contains m features and n samples in total; use the K-SVD algorithm to generate an initial dictionary, and the number of atoms contained in the dictionary and the dimension of each atom in the dictionary are determined according to the characteristics and requirements of the data, and learn a dictionary matrix D ∈ R m ×κ , where R is the set of all possible values of each atom in the dictionary matrix D, and k is the number of basis vectors of the dictionary;

[0134] Find a sparse coefficient matrix B ∈ R k×n such that each sample x i can be represented as a linear combination of dictionary vectors, as shown in the following formula:

[0135] x i ≈ Db i

[0136] where b i is the sparse coding of the sample x i ;

[0137] Iteratively perform sparse coding and dictionary update until convergence or reaching the preset number of iterations, obtain the updated dictionary D′ and the sparse coefficient matrix B′, reconstruct the feature data matrix G, obtain the reconstructed feature data matrix G′, and evaluate the effect of the sparse coding algorithm through the reconstruction error to verify the effectiveness of the sparse coding algorithm;

[0138] Use the sparse coding algorithm to reconstruct the feature data matrix G to obtain the reconstructed feature data matrix G′, and classify the faults of the GIS device into different types, including contact electrical contact faults, contact partial discharge faults, pot insulator mechanical faults, pot insulator breakdown faults, and pot insulator flashover faults;

[0139] Step 4.2: Comprehensively judge the fault information of GIS equipment by combining vibration characteristics and partial discharge characteristics, including fault feature matching and fault location;

[0140] Input the vibration characteristics and partial discharge characteristics data to be detected into the SVM model for fault feature matching to obtain fault feature matching data, and use the sparse coding algorithm adopted in Step 4.1 to output the category labels of the corresponding fault types;

[0141] Input the vibration characteristics and partial discharge characteristics data to be detected into the EKF algorithm for fault location. In this process, by establishing a fault propagation model of GIS equipment to describe the propagation characteristics of vibration signals and partial discharge signals and the changes during fault occurrence, use the EKF algorithm to process the vibration characteristics and partial discharge characteristics data, and obtain the estimated value of the fault location by iteratively updating the estimated value of the state variable;

[0142] Judge the fault result of GIS equipment according to the fault feature matching data and the fault location data, store the diagnosed fault information in the data storage and alarm device, and issue a fault alarm;

[0143] Step 4.3: Use the reverse process interactive modeling method to reconstruct and output the texture and geometric information of GIS equipment, establish a three-dimensional digital twin visualization operation model of GIS equipment, and realize visual monitoring of the operation of GIS equipment in combination with fault information;

[0144] Establish a three-dimensional digital twin visualization model of GIS equipment, and realize visual monitoring of the operation of GIS equipment in combination with fault information, as Figure 2 shown.

[0145] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope defined by the claims of the present invention.

Claims

1. A GIS equipment detection system based on partial discharge-vibration signal fusion, characterized in that: It includes a vibration signal acquisition module, a partial discharge signal acquisition module, a vibration signal feature analysis module, a partial discharge signal feature analysis module and a fault diagnosis and monitoring module; The vibration signal acquisition module is used to collect vibration signals of GIS equipment. The vibration signals of GIS equipment are collected by connecting the contacts of multiple vibration sensors to different positions to be measured of GIS equipment, pre-processing the collected vibration signals of GIS equipment, and obtaining independent vibration data collected at the contacts of each vibration sensor and the basin insulator respectively through a blind source separation algorithm, and transmitting the data to the vibration signal feature analysis module; The partial discharge signal acquisition module is used to collect partial discharge signals of GIS equipment. The partial discharge signals of GIS equipment are collected by connecting the contacts of multiple partial discharge sensors to different positions to be tested of GIS equipment, pre-processing the collected partial discharge signals of GIS equipment, and obtaining independent partial discharge data collected at the contacts of each partial discharge sensor and the basin insulator through a blind source separation algorithm, and transmitting the data to the partial discharge signal feature analysis module. The vibration signal characteristic analysis module is used to calculate and extract the vibration characteristics of the GIS equipment, and transmit the vibration characteristic data to the fault diagnosis and monitoring module; The partial discharge signal characteristic analysis module is used to convert different types of partial discharge signals from analog signals to digital signals, calculate and extract partial discharge characteristics of GIS equipment, and transmit partial discharge characteristic data to the fault diagnosis and monitoring module; The fault diagnosis and monitoring module is used to identify and locate the fault type and location information of the GIS equipment contained in the characteristic data obtained by the vibration characteristic analysis module and the partial discharge characteristic analysis module, store the identified fault information and issue a fault alarm, and establish a three-dimensional digital twin visualization model of the GIS equipment to realize the operation monitoring of the GIS equipment.

2. The GIS equipment detection system based on partial discharge-vibration signal fusion according to claim 1 is characterized in that: The vibration signal acquisition module includes a multi-channel vibration sensor sequence composed of multiple vibration sensors, a vibration signal data preprocessing submodule and a first EMD-ICA blind source separation submodule. The multi-channel vibration sensor sequence includes multiple vibration sensors installed at different test positions of the GIS equipment, and the contacts of the multiple vibration sensors are connected to different test positions of the GIS equipment. The vibration signal data preprocessing submodule is used to improve the quality of the vibration signal and ensure that the data between different vibration sensors have a consistent scale. The first EMD-ICA blind source separation submodule is used to separate the vibration data of the vibration sensor contacts and the independent vibration data of the basin insulator collected by each vibration sensor, and transmit them to the vibration signal feature analysis module.

3. The GIS equipment detection system based on partial discharge-vibration signal fusion according to claim 2 is characterized in that: The partial discharge signal acquisition module includes a multi-channel partial discharge sensor sequence composed of multiple partial discharge sensors, a partial discharge signal data preprocessing submodule and a second EMD-ICA blind source separation submodule. The multi-channel partial discharge sensor sequence includes multiple partial discharge sensors installed at different test positions of the GIS equipment, and the contacts of the multiple partial discharge sensors are connected to different test positions of the GIS equipment. The partial discharge signal data preprocessing submodule is used to improve the quality of the partial discharge signal and ensure that the data between different partial discharge sensors have a consistent scale. The second EMD-ICA blind source separation submodule is used to separate the partial discharge data of the partial discharge sensor contacts and the independent partial discharge data of the basin insulator collected by each partial discharge sensor, and transmit them to the partial discharge signal feature analysis module.

4. The GIS equipment detection system based on partial discharge-vibration signal fusion according to claim 3 is characterized in that: The fault diagnosis and monitoring module includes a fault diagnosis submodule, a data storage and alarm device, and a three-dimensional digital twin visualization operation model construction submodule. The fault diagnosis submodule is used to identify and locate the fault type and location information of the GIS equipment contained in the characteristic data obtained by the vibration characteristic analysis module and the partial discharge characteristic analysis module. The data storage and alarm device is used to store fault information and issue a fault alarm. The three-dimensional digital twin visualization operation model construction submodule is used to generate a GIS equipment visualization model.

5. A GIS equipment detection method based on partial discharge-vibration signal fusion, which performs GIS equipment fault detection based on the system described in claim 1, characterized in that: The following steps are involved: Step 1: The vibration signal acquisition module and the partial discharge signal acquisition module acquire and separate independent vibration signals and independent partial discharge signals of different test positions of the GIS equipment; Step 2: The vibration signal feature extraction module uses the MFE algorithm to calculate and extract the vibration features of independent vibration signals at the contacts and the pot insulators at different test positions of the GIS equipment; Step 3: The partial discharge signal feature extraction module uses the MFE algorithm to calculate and extract the partial discharge features of the independent partial discharge signals at the contacts and the pot insulators at different test positions of the GIS equipment; Step 4: The fault diagnosis and monitoring module classifies the fault information contained in the GIS equipment according to the vibration characteristics and partial discharge characteristics, identifies the fault type and location of the GIS equipment, establishes a three-dimensional digital twin visualization model of the GIS equipment, and realizes visual monitoring of the operation of the GIS equipment in combination with the fault information.

6. The GIS equipment detection method based on partial discharge-vibration signal fusion according to claim 5 is characterized in that: The step 1 specifically includes: Step 1.1, according to the requirements of the GIS equipment detection task, multiple vibration sensors in the multi-channel vibration signal sensor sequence and multiple partial discharge sensors in the multi-channel partial discharge sensor sequence are arranged at different test positions of the GIS equipment to comprehensively collect vibration signals and partial discharge signals of different test positions of the GIS equipment, including vibration signals at the vibration sensor contacts, vibration signals at the pot insulators, partial discharge signals at the partial discharge sensor contacts, and partial discharge signals at the pot insulators; Step 1.2: Preprocess the collected GIS equipment vibration signals and partial discharge signals to improve signal quality and ensure that the data between different samples have a consistent scale; Step 1.3: Use the EMD-ICA blind source separation algorithm to separate and screen out the independent vibration signals at each vibration sensor contact and basin insulator, and the independent partial discharge signals at each partial discharge sensor contact and basin insulator, and extract the initial features of the vibration signals at the vibration sensor contacts and basin insulators of the GIS equipment, and the initial features of the partial discharge signals at the partial discharge sensor contacts and basin insulators.

7. The GIS equipment detection method based on partial discharge-vibration signal fusion according to claim 6 is characterized in that: The step 2 specifically includes: Step 2.1, coarse-graining the vibration signal after blind source separation, and dividing the vibration signal into several segments using the time scale factor; Vibration signal y after coarse-grained blind source separation i (t), use the time scale factor τ to divide the vibration signal into several segments, let 1≤j≤N, and convert y i Divide into τ segments, each segment length is N / τ, as shown in the following formula: Step 2.2, using the MFE algorithm to calculate the fuzzy entropy of each vibration signal, and selecting the component with the smallest entropy as the vibration feature of the GIS equipment; Select different time scale factors τ to calculate the fuzzy entropy FuzzyEn of the feature signal. Given the embedding dimension c, similarity tolerance r and fuzzy function gradient grad, the fuzzy entropy of each coarse-grained signal is calculated as shown in the following formula: The component with the smallest entropy is selected as the vibration feature p.

8. The GIS equipment detection method based on partial discharge-vibration signal fusion according to claim 7 is characterized in that: Step 3 specifically includes: Step 3.1, converting different types of partial discharge signals from analog signals to digital signals as partial discharge characteristic signals through an analog-to-digital converter; Step 3.2: Use the same method as step 2 to extract the vibration characteristics of the GIS equipment to extract the partial discharge characteristic q of the partial discharge characteristic signal.

9. The GIS equipment detection method based on partial discharge-vibration signal fusion according to claim 8 is characterized in that: Step 4 specifically includes: Step 4.1, based on the vibration characteristics and partial discharge characteristics obtained by the vibration signal characteristic analysis module and the partial discharge signal characteristic analysis module, a sparse coding algorithm is used to classify the faults of the GIS equipment; The feature data matrix G is constructed using the vibration feature p and the partial discharge feature q, where each column represents a sample and each row represents a feature, containing a total of m features and n samples. The K-SVD algorithm is used to generate the initial dictionary. The number of atoms contained in the dictionary and the dimension of each atom in the dictionary are determined according to the characteristics and requirements of the data. A dictionary matrix D∈R is learned. m×κ , where R is the set of all possible values ​​of each atom in the dictionary matrix D, and k is the number of basis vectors of the dictionary; Find a sparse coefficient matrix B∈R k×n So that each sample x i You can use a linear combination of dictionary vectors, as shown in the following formula: x i ≈Db i Among them, b i For sample x i Sparse coding; Iterate sparse coding and dictionary update until convergence or reaching the preset number of iterations, obtain the updated dictionary D′ and sparse coefficient matrix B′, reconstruct the feature data matrix G, and obtain the reconstructed feature data matrix G′. The effect of the sparse coding algorithm is evaluated by the reconstruction error to verify the effectiveness of the sparse coding algorithm. The sparse coding algorithm is used to reconstruct the characteristic data matrix G to obtain the reconstructed characteristic data matrix G′, which can classify the GIS equipment faults into different types, including contact electrical contact fault, contact partial discharge fault, pot insulator mechanical fault, pot insulator breakdown fault, and pot insulator flashover fault. Step 4.2: Comprehensively judge the GIS equipment fault information by combining the vibration characteristics and partial discharge characteristics, including fault feature matching and fault location; Input the vibration characteristics and partial discharge characteristics data to be detected into the SVM model for fault feature matching, obtain fault feature matching data, and use the sparse coding algorithm used in step 4.1 to output the category label of the corresponding fault type; The vibration characteristics and partial discharge characteristics data to be detected are input into the EKF algorithm for fault location. The fault propagation model of GIS equipment is established to describe the propagation characteristics of vibration signals and partial discharge signals and the changes when the fault occurs. The vibration characteristics and partial discharge characteristics data are processed using the EKF algorithm. The estimated value of the fault location is obtained by iteratively updating the estimated value of the state variable. According to the fault feature matching data and fault location data, the fault result of the GIS equipment is judged, the diagnosed fault information is stored in the data storage and alarm device, and a fault alarm is issued; Step 4.3: Use an interactive inverse process modeling method to reconstruct and output the texture and geometric information of the GIS equipment, establish a three-dimensional digital twin visualization operation model of the GIS equipment, and combine fault information to realize visual monitoring of the operation of the GIS equipment.

Images