Method and system for detecting and positioning partial discharge of switch cabinet

Abstract

The invention relates to the technical field of power monitoring, in particular to a method and system for detecting and positioning partial discharge of a switch cabinet, and the system comprises a signal collection module, a signal preprocessing module, a feature extraction module, a data fusion module, a discharge source positioning module and a classification and alarm module. The signal acquisition module simultaneously acquires partial discharge signals through a plurality of sensors installed in the switch cabinet, the signal preprocessing module performs noise filtering and signal enhancement on the acquired signals, and the feature extraction module screens out most representative discharge features based on a deep learning algorithm. The data fusion module fuses data of multiple sensors through Kalman filtering and a Bayesian reasoning algorithm, the discharge source positioning module accurately calculates the position of a discharge source through a multi-point positioning algorithm, and the discharge source is classified through the classification and alarm module. The partial discharge detection precision and the positioning accuracy of the switch cabinet can be effectively improved, and safe operation of a power system is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of power monitoring, and in particular to a detection and positioning method and system for partial discharge of a switch cabinet. Background Art

[0002] With the continuous development of power systems, switchgear, as an important equipment in the power grid, is widely used in various power transmission and distribution systems. The insulation and connection components in the switchgear are prone to partial discharge under high voltage environment. Although the initial discharge intensity of partial discharge is small, long-term accumulation will cause irreversible damage to the insulation material, which will lead to equipment failure and even serious power accidents. Therefore, the effective detection and positioning of partial discharge inside the switchgear is of great significance to ensure the safe operation of the power system.

[0003] At present, the existing partial discharge detection technology has the following main problems: first, the limitation of single sensor signal acquisition leads to limited coverage of detection signal, which is prone to missed detection or false detection; second, the signal processing method is not sufficient to effectively filter out noise interference in complex environments, resulting in low accuracy and robustness of detection signals; finally, traditional discharge source location and classification methods lack intelligence and accuracy, and cannot effectively identify and locate local discharge sources in complex electromagnetic environments. Summary of the invention

[0004] Based on the above purpose, the present invention provides a method and system for detecting and locating partial discharge of a switch cabinet.

[0005] The detection and positioning method of partial discharge of a switch cabinet comprises the following steps:

[0006] S1, obtaining the original signal: collecting the partial discharge signal simultaneously through multiple sensors installed inside the switch cabinet, wherein the sensors include a high-frequency current sensor, an ultrasonic sensor and an ultra-high frequency electromagnetic wave sensor;

[0007] S2, signal preprocessing: preliminary noise filtering and signal enhancement are performed on the collected partial discharge signals. The adaptive filtering algorithm is combined with wavelet transform to perform multi-scale decomposition of the signal to remove power frequency noise and environmental electromagnetic interference, while retaining the time and frequency information of the signal;

[0008] S3, feature extraction: extract features from the preprocessed signal. The extracted features include the amplitude, phase, frequency, duration and energy of the discharge signal. The features are screened using a feature selection algorithm based on deep learning to extract several features that best represent the local discharge characteristics.

[0009] S4, multi-sensor data fusion: Based on Kalman filtering and Bayesian reasoning algorithm, the data of different sensors are fused to improve the robustness and accuracy of the detection signal, and the global partial discharge feature vector is obtained through data fusion;

[0010] S5, discharge source location: using a multi-point positioning algorithm, combined with the time difference of the received signal, amplitude attenuation, and the spatial layout of the sensor, the precise location of the local discharge source is calculated;

[0011] S6, discharge source classification and alarm: Based on the extracted feature vector and positioning information, the support vector machine is used to classify the local discharge source, and an alarm signal is issued in combination with the classification result and real-time detection data.

[0012] Optionally, the S1 specifically includes:

[0013] High-frequency current sensor: installed at the cable entrance and key electrical contacts of the switch cabinet to collect high-frequency current signals caused by partial discharge;

[0014] Ultrasonic sensor: Installed on the inner wall of the switch cabinet, close to the insulating material or electrical connection where partial discharge may occur, to capture the ultrasonic signal generated by partial discharge;

[0015] Ultra-high frequency electromagnetic wave sensors: arranged in multiple corners inside the switch cabinet, especially in the upper part of the switch cabinet and near key electrical components, to capture the ultra-high frequency electromagnetic wave signals generated by partial discharge.

[0016] Optionally, the S2 specifically includes:

[0017] S21, noise filtering: the filter parameters are adjusted according to the characteristics of the real-time signal through the adaptive filter to effectively remove the power frequency noise and low-frequency interference signals;

[0018] S22, wavelet transform: perform multi-scale wavelet transform on the filtered signal to decompose the signal into components of different scales, extract the effective discharge characteristics in the signal, and filter out environmental electromagnetic interference;

[0019] S23, signal enhancement: enhance the signal at each wavelet scale, retain the high-frequency feature information, and restore the signal in combination with inverse transformation.

[0020] Optionally, the S3 specifically includes:

[0021] S31, feature extraction: extracting features from the preprocessed signal;

[0022] S32, feature selection: a convolutional neural network is used to screen the extracted feature vectors, and the feature selection is gradually optimized through the forward propagation and back propagation process of the network, and finally several features that best represent the local discharge characteristics are selected;

[0023] S33, model training: The convolutional neural network model is trained using the labeled partial discharge dataset so that it can automatically select the most representative features.

[0024] Optionally, the S4 specifically includes:

[0025] S41, Kalman filtering: preliminarily filters the data collected by each sensor, corrects the influence of noise, predicts the signal state at the next moment, and integrates the actual measurement value at the current moment to achieve smooth fusion of signals;

[0026] S42, Bayesian reasoning: Based on Kalman filtering, the Bayesian reasoning method is used to perform probabilistic fusion of multi-sensor data, combining prior information and observed data to generate a global local discharge feature vector.

[0027] Optionally, the S5 specifically includes:

[0028] S51, time difference method: calculating the position of the discharge source based on the time difference of the discharge signals received by multiple sensors;

[0029] S52, amplitude attenuation method: combining the attenuation degree of the amplitude of the signal received by the sensor, estimating the location of the discharge source, and performing triangulation positioning through the amplitude information of multiple sensors;

[0030] S53, sensor spatial layout correction: Consider the spatial layout of the sensor, correct the positioning error caused by the environment or equipment structure, and obtain the precise location of the discharge source.

[0031] Optionally, the S6 specifically includes:

[0032] S61, classification model training: using the labeled partial discharge data, training the support vector machine model, the model uses the radial basis kernel function, with the goal of maximizing the classification interval, and generates the classification decision surface;

[0033] S62, real-time classification: inputting the partial discharge feature vector detected in real time into the trained support vector machine model to identify the type of discharge source and determine the danger level of the discharge source;

[0034] S63, alarm mechanism: based on the classification results and real-time detection data, when the system detects a dangerous level of partial discharge, it automatically triggers an audible and visual alarm and sends the alarm information to the monitoring system.

[0035] A partial discharge detection and positioning system for a switch cabinet, used to implement the above-mentioned partial discharge detection and positioning method for a switch cabinet, comprises the following modules:

[0036] Signal acquisition module: equipped with multiple sensors installed inside the switch cabinet, used to simultaneously collect partial discharge signals, the sensors include high-frequency current sensors, ultrasonic sensors and ultra-high frequency electromagnetic wave sensors;

[0037] Signal preprocessing module: used to filter out noise and enhance the collected partial discharge signal, use adaptive filtering algorithm combined with wavelet transform to decompose the signal at multiple scales, remove power frequency noise and environmental electromagnetic interference, and retain the time and frequency information of the signal;

[0038] Feature extraction module: used to extract features from the preprocessed signal. The extracted features include the amplitude, phase, frequency, duration and energy of the discharge signal. The feature selection algorithm based on deep learning is used to screen the features and extract several features that best represent the local discharge characteristics.

[0039] Data fusion module: Based on Kalman filtering and Bayesian inference algorithm, data from different sensors are fused to improve the robustness and accuracy of detection signals, and global partial discharge feature vectors are obtained through data fusion;

[0040] Discharge source location module: uses a multi-point location algorithm, combined with the time difference of the received signal, amplitude attenuation, and the spatial layout of the sensor, to calculate the precise location of the local discharge source;

[0041] Classification and alarm module: Based on the extracted feature vectors and positioning information, the support vector machine is used to classify the local discharge source, and an alarm signal is issued in combination with the classification results and real-time detection data.

[0042] Beneficial effects of the present invention:

[0043] The present invention uses a high-frequency current sensor, an ultrasonic sensor and an ultra-high frequency electromagnetic wave sensor in combination, so that the system can simultaneously collect different types of signals generated by partial discharge. The sensors are arranged at key locations of the switch cabinet to ensure that partial discharge can be detected in a timely manner no matter where it occurs. The coordinated work of multiple sensors not only improves the accuracy of signal acquisition, but also expands the monitoring range, effectively avoiding missed detection of discharge sources, and providing high-quality data input for subsequent signal processing and positioning.

[0044] The present invention adopts a signal preprocessing method combining an adaptive filtering algorithm and wavelet transform, which can effectively remove noise interference and enhance the significance of key discharge signals. Combining the multi-sensor data fusion technology of Kalman filtering and Bayesian reasoning, the system can still accurately extract global local discharge characteristics in a noisy environment. This multi-level signal processing and data fusion not only enhances the anti-interference ability of the system, but also greatly improves the accuracy and robustness of the detection results, laying the foundation for the precise positioning of the discharge source.

[0045] The present invention uses a multi-point positioning algorithm to accurately calculate the precise location of the local discharge source, and combines the classification function of the support vector machine to effectively classify the discharge source. When the system detects a discharge source of a dangerous level, it can promptly send out an alarm signal to remind maintenance personnel to quickly take corresponding protective measures. This intelligent positioning and classification function can not only significantly shorten the fault location time, but also improve the safety and operating efficiency of power equipment, and effectively prevent the expansion or escalation of potential faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0047] Figure 1 A schematic diagram of a method flow of an embodiment of the present invention;

[0048] Figure 2 Schematic diagram of a system flow of an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0050] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0051] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0052] like Figure 1 As shown, the detection and positioning method of partial discharge of a switch cabinet comprises the following steps:

[0053] S1, obtaining the original signal: the partial discharge signal is collected simultaneously by multiple sensors installed inside the switch cabinet, including high-frequency current sensors, ultrasonic sensors and ultra-high frequency electromagnetic wave sensors;

[0054] S2, signal preprocessing: preliminary noise filtering and signal enhancement are performed on the collected partial discharge signals. The adaptive filtering algorithm is combined with wavelet transform to perform multi-scale decomposition of the signal to remove power frequency noise and environmental electromagnetic interference, while retaining the time and frequency information of the signal;

[0055] S3, feature extraction: extract features from the preprocessed signal. The extracted features include the amplitude, phase, frequency, duration and energy of the discharge signal. The features are screened using a feature selection algorithm based on deep learning to extract several features that best represent the local discharge characteristics.

[0056] S4, multi-sensor data fusion: Based on Kalman filtering and Bayesian reasoning algorithm, the data of different sensors are fused to improve the robustness and accuracy of the detection signal, and the global partial discharge feature vector is obtained through data fusion;

[0057] S5, discharge source location: using a multi-point positioning algorithm, combined with the time difference of the received signal, amplitude attenuation, and the spatial layout of the sensor, the precise location of the local discharge source is calculated;

[0058] S6, discharge source classification and alarm: Based on the extracted feature vector and positioning information, the support vector machine is used to classify the local discharge source, and an alarm signal is issued in combination with the classification result and real-time detection data.

[0059] S1 specifically includes:

[0060] High-frequency current sensor: installed at the cable entrance and key electrical contacts of the switch cabinet to collect high-frequency current signals caused by partial discharge;

[0061] Ultrasonic sensor: Installed on the inner wall of the switch cabinet, close to the insulating material or electrical connection where partial discharge may occur, to capture the ultrasonic signal generated by partial discharge;

[0062] Ultra-high frequency electromagnetic wave sensors: arranged in multiple corners inside the switch cabinet, especially in the upper part of the switch cabinet and near key electrical components, to capture ultra-high frequency electromagnetic wave signals generated by partial discharge;

[0063] By rationally arranging different types of sensors, it is possible to comprehensively collect partial discharge signals at different locations in the switch cabinet, providing a reliable data source for subsequent signal processing and positioning.

[0064] S2 specifically includes:

[0065] S21, noise filtering: the filter parameters are adjusted according to the characteristics of the real-time signal through the adaptive filter to effectively remove the power frequency noise and low-frequency interference signals;

[0066] S22, wavelet transform: perform multi-scale wavelet transform on the filtered signal to decompose the signal into components of different scales, extract the effective discharge characteristics in the signal, and filter out environmental electromagnetic interference;

[0067] S23, signal enhancement: enhance the signal at each wavelet scale, retain the high-frequency feature information, and restore the signal in combination with inverse transformation;

[0068] The preprocessing method combining adaptive filtering and wavelet transform can effectively filter out noise, enhance the key features in the signal, and provide high-quality signals for subsequent feature extraction.

[0069] S3 specifically includes:

[0070] S31, feature extraction: extracting features from the preprocessed signal;

[0071] S32, feature selection: a convolutional neural network is used to screen the extracted feature vectors, and the feature selection is gradually optimized through the forward propagation and back propagation process of the network, and finally several features that best represent the local discharge characteristics are selected;

[0072] S33, model training: Use the labeled partial discharge dataset to train the convolutional neural network model so that it can automatically select the most representative features;

[0073] Through the feature selection algorithm based on deep learning, the system can automatically screen out the features with the most diagnostic value from a large number of signal features, thereby improving the accuracy of partial discharge detection.

[0074] S4 specifically includes:

[0075] S41, Kalman filtering: preliminarily filters the data collected by each sensor, corrects the influence of noise, predicts the signal state at the next moment, and integrates the actual measurement value at the current moment to achieve smooth fusion of signals;

[0076] S42, Bayesian reasoning: Based on Kalman filtering, the Bayesian reasoning method is used to perform probabilistic fusion of multi-sensor data, combining prior information and observed data to generate a global partial discharge feature vector;

[0077] Through the multi-sensor data fusion method combining Kalman filtering and Bayesian reasoning, the system can obtain more accurate and robust partial discharge detection signals in noisy environments.

[0078] S5 specifically includes:

[0079] S51, time difference method: based on the time difference of multiple sensors receiving discharge signals, the position of the discharge source is calculated as:

[0080]

[0081] Where d is the distance between the discharge source and the sensor, Δt is the time difference between the two sensors receiving the signal, and v is the propagation speed of the electromagnetic wave in the switch cabinet;

[0082] S52, amplitude attenuation method: combining the attenuation degree of the amplitude of the signal received by the sensor, estimating the location of the discharge source, and performing triangulation positioning through the amplitude information of multiple sensors;

[0083] S53, sensor spatial layout correction: considering the spatial layout of the sensor, correcting the positioning error caused by the environment or equipment structure, and obtaining the precise location of the discharge source;

[0084] By combining the time difference method, amplitude attenuation method and spatial layout correction, the location of the local discharge source can be accurately calculated, providing precise positioning information for fault troubleshooting.

[0085] S6 specifically includes:

[0086] S61, classification model training: using the labeled partial discharge data, training the support vector machine model, the model uses the radial basis kernel function, with the goal of maximizing the classification interval, and generates the classification decision surface;

[0087] S62, real-time classification: inputting the partial discharge feature vector detected in real time into the trained support vector machine model to identify the type of discharge source and determine the danger level of the discharge source;

[0088] S63, alarm mechanism: based on the classification results and real-time detection data, when the system detects a dangerous level of partial discharge, it automatically triggers an audible and visual alarm and sends the alarm information to the monitoring system;

[0089] By training the support vector machine model to classify and alarm the discharge source, the system can accurately identify different types of partial discharges and issue an alarm in a timely manner when potential dangers are detected, ensuring the safe operation of the equipment.

[0090] like Figure 2 As shown, a partial discharge detection and positioning system of a switch cabinet is used to implement the above-mentioned partial discharge detection and positioning method of a switch cabinet, and includes the following modules:

[0091] Signal acquisition module: equipped with multiple sensors installed inside the switch cabinet, used to simultaneously collect partial discharge signals, the sensors include high-frequency current sensors, ultrasonic sensors and ultra-high frequency electromagnetic wave sensors;

[0092] Signal preprocessing module: used to filter out noise and enhance the collected partial discharge signal, use adaptive filtering algorithm combined with wavelet transform to decompose the signal at multiple scales, remove power frequency noise and environmental electromagnetic interference, and retain the time and frequency information of the signal;

[0093] Feature extraction module: used to extract features from the preprocessed signal. The extracted features include the amplitude, phase, frequency, duration and energy of the discharge signal. The feature selection algorithm based on deep learning is used to screen the features and extract several features that best represent the local discharge characteristics.

[0094] Data fusion module: Based on Kalman filtering and Bayesian inference algorithm, data from different sensors are fused to improve the robustness and accuracy of detection signals, and global partial discharge feature vectors are obtained through data fusion;

[0095] Discharge source location module: uses a multi-point location algorithm, combined with the time difference of the received signal, amplitude attenuation, and the spatial layout of the sensor, to calculate the precise location of the local discharge source;

[0096] Classification and alarm module: Based on the extracted feature vectors and positioning information, the support vector machine is used to classify the local discharge source, and an alarm signal is issued in combination with the classification results and real-time detection data.

[0097] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0098] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for detecting and locating partial discharge of a switch cabinet, characterized in that: The following steps are involved: S1, obtaining the original signal: collecting the partial discharge signal simultaneously through multiple sensors installed inside the switch cabinet, wherein the sensors include a high-frequency current sensor, an ultrasonic sensor and an ultra-high frequency electromagnetic wave sensor; S2, signal preprocessing: preliminary noise filtering and signal enhancement are performed on the collected partial discharge signals. The adaptive filtering algorithm is combined with wavelet transform to perform multi-scale decomposition of the signal to remove power frequency noise and environmental electromagnetic interference, while retaining the time and frequency information of the signal; S3, feature extraction: extract features from the preprocessed signal. The extracted features include the amplitude, phase, frequency, duration and energy of the discharge signal. The features are screened using a feature selection algorithm based on deep learning to extract several features that best represent the local discharge characteristics. S4, multi-sensor data fusion: Based on Kalman filtering and Bayesian reasoning algorithm, the data of different sensors are fused to improve the robustness and accuracy of the detection signal, and the global partial discharge feature vector is obtained through data fusion; S5, discharge source location: using a multi-point positioning algorithm, combined with the time difference of the received signal, amplitude attenuation, and the spatial layout of the sensor, the precise location of the local discharge source is calculated; S6, discharge source classification and alarm: Based on the extracted feature vector and positioning information, the support vector machine is used to classify the local discharge source, and an alarm signal is issued in combination with the classification result and real-time detection data.

2. The method for detecting and locating partial discharge of a switch cabinet according to claim 1, characterized in that: The S1 specifically includes: High-frequency current sensor: used to collect high-frequency current signals caused by partial discharge; Ultrasonic sensor: used to capture ultrasonic signals generated by partial discharge; Ultra-high frequency electromagnetic wave sensor: used to capture ultra-high frequency electromagnetic wave signals generated by partial discharge.

3. The method for detecting and locating partial discharge of a switch cabinet according to claim 2, characterized in that: The S2 specifically includes: S21, noise filtering: the filter parameters are adjusted according to the characteristics of the real-time signal through the adaptive filter to effectively remove the power frequency noise and low-frequency interference signals; S22, wavelet transform: perform multi-scale wavelet transform on the filtered signal to decompose the signal into components of different scales, extract the effective discharge characteristics in the signal, and filter out environmental electromagnetic interference; S23, signal enhancement: enhance the signal at each wavelet scale, retain the high-frequency feature information, and restore the signal in combination with inverse transformation.

4. The method for detecting and locating partial discharge of a switch cabinet according to claim 3, characterized in that: The S3 specifically includes: S31, feature extraction: extracting features from the preprocessed signal; S32, feature selection: a convolutional neural network is used to screen the extracted feature vectors, and the feature selection is gradually optimized through the forward propagation and back propagation process of the network, and finally several features that best represent the local discharge characteristics are selected; S33, model training: The convolutional neural network model is trained using the labeled partial discharge dataset so that it can automatically select the most representative features.

5. The method for detecting and locating partial discharge of a switch cabinet according to claim 4, characterized in that: The S4 specifically includes: S41, Kalman filtering: preliminarily filters the data collected by each sensor, corrects the influence of noise, predicts the signal state at the next moment, and integrates the actual measurement value at the current moment to achieve smooth fusion of signals; S42, Bayesian reasoning: Based on Kalman filtering, the Bayesian reasoning method is used to perform probabilistic fusion of multi-sensor data, combining prior information and observed data to generate a global local discharge feature vector.

6. The method for detecting and locating partial discharge of a switch cabinet according to claim 5, characterized in that: The S5 specifically includes: S51, time difference method: calculating the position of the discharge source based on the time difference of the discharge signals received by multiple sensors; S52, amplitude attenuation method: combining the attenuation degree of the amplitude of the signal received by the sensor, estimating the location of the discharge source, and performing triangulation positioning through the amplitude information of multiple sensors; S53, sensor spatial layout correction: Consider the spatial layout of the sensor, correct the positioning error caused by the environment or equipment structure, and obtain the precise location of the discharge source.

7. The method for detecting and locating partial discharge of a switch cabinet according to claim 6, characterized in that: The S6 specifically includes: S61, classification model training: using the labeled partial discharge data, training the support vector machine model, the model uses the radial basis kernel function, with the goal of maximizing the classification interval, and generates the classification decision surface; S62, real-time classification: inputting the partial discharge feature vector detected in real time into the trained support vector machine model to identify the type of discharge source and determine the danger level of the discharge source; S63, alarm mechanism: based on the classification results and real-time detection data, when the system detects a dangerous level of partial discharge, it automatically triggers an audible and visual alarm and sends the alarm information to the monitoring system.

8. A detection and positioning system for partial discharge of a switch cabinet, used to implement a detection and positioning method for partial discharge of a switch cabinet as claimed in any one of claims 1 to 9, characterized in that: Includes the following modules: Signal acquisition module: equipped with multiple sensors installed inside the switch cabinet, used to simultaneously collect partial discharge signals, the sensors include high-frequency current sensors, ultrasonic sensors and ultra-high frequency electromagnetic wave sensors; Signal preprocessing module: used to filter out noise and enhance the collected partial discharge signal, use adaptive filtering algorithm combined with wavelet transform to decompose the signal at multiple scales, remove power frequency noise and environmental electromagnetic interference, and retain the time and frequency information of the signal; Feature extraction module: used to extract features from the preprocessed signal. The extracted features include the amplitude, phase, frequency, duration and energy of the discharge signal. The feature selection algorithm based on deep learning is used to screen the features and extract several features that best represent the local discharge characteristics. Data fusion module: Based on Kalman filtering and Bayesian inference algorithm, data from different sensors are fused to improve the robustness and accuracy of detection signals, and global partial discharge feature vectors are obtained through data fusion; Discharge source location module: uses a multi-point location algorithm, combined with the time difference of the received signal, amplitude attenuation, and the spatial layout of the sensor, to calculate the precise location of the local discharge source; Classification and alarm module: Based on the extracted feature vectors and positioning information, the support vector machine is used to classify the local discharge source, and an alarm signal is issued in combination with the classification results and real-time detection data.

Images