A method and device for monitoring high-voltage cable TMR high-frequency partial discharge pulse signal
Through the TMR high-frequency current sensor and multi-scale neural network model, the problem of identifying various local discharge types in high-voltage cables is solved, and the accuracy of local discharge pulse signals of high-voltage cables is realized, which improves the recognition accuracy and safety, and is integrated , low power consumption.
Patent Information
- Application Number
- CN202410292450.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-14
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2044-03-14
AI Technical Summary
The prior art is difficult to accurately distinguish the types of local discharges caused by multiple external factors in high-voltage cables, affecting the accuracy of identification.
The TMR high-frequency current sensor is used to sense the local discharge pulse current signal, and the adaptive wavelet threshold denoising process is performed through the signal processing module to generate a phase distribution map of the local discharge pulse, and a multi-scale neural network model is used to identify the discharge type.
It realizes accurate identification of local discharge pulse signals of high-voltage cables, improves identification accuracy and safety, and has the characteristics of integration and low power consumption.
Smart Images

Figure CN118169523B_ABST
Abstract
Description
[Technical field]
[0001] The present invention relates to the technical field of electric power equipment, and in particular to a method and device for monitoring a high-voltage cable TMR high-frequency partial discharge pulse signal. [Background technology]
[0002] Tunnel magnetoresistance effect (TMR), TMR sensor is the fourth generation of magnetic sensor application technology.
[0003] With the increase in the number of high-voltage cable systems and the demand for power supply stability, it has become an important task to reduce the frequency and duration of power outages to ensure the quality of power supply. In order to achieve this goal, the insulation condition of the cable must be monitored and obtained in real time, which involves collecting a large amount of online detection data and evaluating the operation of the cable. During cable operation, they are subjected to various forms of stress, which may cause partial discharge. Partial discharge is a key factor that causes the degradation of the insulation performance of cables and their joints. If it is not controlled, it may eventually lead to cable insulation failure and cause power outages. Therefore, it is crucial to perform partial discharge detection on high-voltage cable insulation and accurately determine the defect type to ensure the safe operation of the cable.
[0004] At present, the detection signal of partial discharge fault of high-voltage cable is mainly measured by pulse current signal through high-frequency current sensor. Pulse current signal has become one of the most used detection signals because of its advantages of simple acquisition method, low interference, and easy analysis of acquired signals. At present, the identification method of partial discharge fault of high-voltage cable is basically to construct partial discharge patterns that can reflect different types of discharge sources, extract effective feature parameters from these patterns, and train appropriate classifiers to finally make a judgment on the type of partial discharge source. However, this kind of method is usually carried out in a single defect system, and it is proved that the type of partial discharge source can be identified when a single partial discharge occurs in the high-voltage cable. However, during the manufacturing and operation process of high-voltage cable, it is affected by multiple external factors at the same time, and there are often multiple types of partial discharge caused by defects at the same time. The measured partial discharge waveform is mostly the sum of these discharge waveforms, which is difficult to distinguish by existing methods, affecting the accuracy of identification.
[0005] The present invention aims at the technical problem of poor accuracy in monitoring the partial discharge waveform of a high-voltage cable, and makes technical improvements to a method and device for monitoring the partial discharge pulse signal of a high-voltage cable. [Summary of the invention]
[0006] The purpose of the present invention is to provide a high-voltage cable partial discharge pulse signal monitoring method which is integrated, has low power consumption, strong safety and accurate identification.
[0007] To achieve the above object, the technical solution adopted by the present invention is a high-voltage cable TMR high-frequency partial discharge pulse signal monitoring method, comprising the following steps:
[0008] S1, using TMR high-frequency current sensor to sense the local discharge pulse current signal released at the grounding wire of the shield layer of the high-voltage cable joint;
[0009] S2, the signal processing module conditions and AD-converts the partial discharge pulse current signal, performs denoising processing using an improved wavelet threshold denoising method with an adaptive threshold function, and generates a partial discharge pulse phase distribution spectrum;
[0010] S3. The server module inputs the partial discharge pulse phase distribution map into the trained multi-scale neural network model to identify the characteristics of the partial discharge pulse current signal and obtain the corresponding partial discharge type.
[0011] Preferably, the threshold function in step S2 has an adaptive improved wavelet threshold denoising method:
[0012] The improved wavelet threshold function in, represents the kth wavelet coefficient at the jth decomposition scale in the wavelet transform, o is the universal threshold of the wavelet coefficient, a is an adjustable parameter, is the optimized wavelet coefficient;
[0013] Improved universal threshold of wavelet coefficients Among them, j is the decomposition scale, N j is the number of high-frequency coefficients in the jth layer;
[0014] By comparing the intensity of the noise content in the filtered partial discharge pulse current signal, it is determined whether the adjustable parameter a of the current improved wavelet threshold function is applicable. If the noise intensity increases, the adjustable parameter a is adaptively adjusted, otherwise it remains unchanged.
[0015] Preferably, step S3: the multi-scale neural network model diagram includes several branch models of different scales, which are divided into two parts: feature extraction and feature connection; first, the feature extraction part, as the network layer increases, the number of feature maps increases exponentially, and the size of the feature maps also decreases accordingly; second, the feature connection part, through the feature extraction of different layers, features of different scales are connected to the cascade layer; finally, through full connection, feature fusion is realized in the output layer, and the recognition result is comprehensively output using the Softmax function.
[0016] Preferably, step S3: the local discharge types include four types: corona discharge, surface discharge, internal discharge and arc discharge.
[0017] Another object of the present invention is to provide an integrated, low-power, highly safe, and accurately identified high-voltage cable partial discharge pulse signal monitoring device.
[0018] To achieve the above-mentioned another purpose, the technical solution adopted by the present invention is a high-voltage cable TMR high-frequency partial discharge pulse signal monitoring device, including a TMR high-frequency current sensor, a field signal receiver and a back-end processor; the field signal receiver is used to receive the partial discharge pulse current signal transmitted by the TMR current sensor through a coaxial cable and transmit it to the back-end processor; the back-end processor includes a signal processing module and a server module; the monitoring device is used to execute a high-voltage cable TMR high-frequency partial discharge pulse signal monitoring method as described in any one of claims 1 to 4.
[0019] Preferably, the high-voltage cable TMR high-frequency partial discharge pulse signal monitoring device comprises a plurality of TMR high-frequency current sensors, some of which are installed at the joints of the first and second sections of the A phase, the third and first sections of the B phase, and the second and third sections of the C phase of the three-phase crossed high-voltage cable of the power grid, and the other part of the TMR high-frequency current sensors are installed at the joints of the second and third sections of the A phase, the first and second sections of the B phase, and the third and first sections of the C phase of the three-phase crossed high-voltage cable of the power grid.
[0020] Preferably, the high-voltage cable TMR high-frequency partial discharge pulse signal monitoring device also includes a WAPI authentication server; the field signal receiver transmits the partial discharge pulse current signal data to the back-end processor through the WAPI private design protocol, and the WAPI authentication server is used to set keys and verification mechanisms for the field signal receiver and the back-end processor, and prevent unauthorized external nodes from intercepting and tampering with the partial discharge pulse current signal data through encryption algorithms and security authentication protocols.
[0021] Preferably, the signal processing module includes a signal conditioning unit, an AD conversion unit, a denoising unit and a spectrum generating unit; the signal conditioning unit includes a charge conversion circuit, an active filtering circuit and an output amplifier circuit, the AD conversion unit is used to convert the output voltage of the partial discharge pulse current signal after signal conditioning into a digital signal and complete automatic acquisition, and the denoising unit is used to denoise the digital signal.
[0022] Preferably, the server module comprises a feature extraction unit and a discharge type identification unit.
[0023] The beneficial effects of a high-voltage cable TMR high-frequency partial discharge pulse signal monitoring method and device of the present invention are as follows: a field signal receiver is used to receive a partial discharge pulse current signal transmitted by a TMR current sensor through a coaxial cable, and transmit the data to a back-end processor through a WAPI private design protocol; the back-end processor mainly includes a signal processing module and a server module, the signal processing module performs conditioning, AD conversion, denoising and generates a partial discharge spectrum for the partial discharge pulse current signal, the server reads the spectrum denoised by the signal processing module, identifies features through a multi-scale network, and obtains the corresponding partial discharge type; the time-space synchronous detection of the high-frequency partial discharge signal of the high-voltage cable can be realized, and the device is highly portable and has the advantages of integration, low power consumption, high safety, accurate identification, etc.
Brief Description of the Drawings
[0024] Figure 1 The present invention is a step-by-step diagram of a method for monitoring high-voltage cable TMR high-frequency partial discharge pulse signals.
[0025] Figure 2 This is a diagram of the architecture of a high-voltage cable TMR high-frequency partial discharge pulse signal monitoring device.
[0026] Figure 3 It is a module diagram of the signal conditioning circuit of the present invention.
[0027] Figure 4 It is a flow chart of the improved wavelet threshold denoising method of the present invention.
[0028] Figure 5 This is a multi-scale neural network model diagram of the present invention. [Specific implementation method]
[0029] The features and exemplary embodiments of various aspects of the present invention will be described in detail below. In the detailed description below, many specific details are proposed to provide a comprehensive understanding of the present invention. But it is obvious to those skilled in the art that the present invention can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present invention by illustrating examples of the present invention. The present invention is by no means limited to any specific configuration and algorithm proposed below, but covers any modification, replacement and improvement of elements, parts and algorithms without departing from the concept of the present invention. In the accompanying drawings and the following description, known structures and technologies are not shown to avoid unnecessary ambiguity to the present invention.
[0030] Example 1
[0031] This embodiment implements a method for monitoring high-voltage cable TMR high-frequency partial discharge pulse signals.
[0032] Figure 1This is a step-by-step diagram of a method for monitoring high-voltage cable TMR high-frequency partial discharge pulse signals. Figure 1 As shown, the method of this embodiment includes the following steps:
[0033] S1, using TMR high-frequency current sensor to sense the local discharge pulse current signal released at the grounding wire of the shield layer of the high-voltage cable joint;
[0034] S2, conditioning and AD conversion of the partial discharge pulse current signal, using an improved wavelet threshold denoising method with adaptive threshold function to perform denoising, and generating a partial discharge pulse phase distribution spectrum;
[0035] S3. Input the partial discharge pulse phase distribution map into the trained multi-scale neural network model to identify the characteristics of the partial discharge pulse current signal and obtain the corresponding partial discharge type.
[0036] Example 2
[0037] This embodiment implements a high-voltage cable TMR high-frequency partial discharge pulse signal monitoring device.
[0038] Figure 2 This is a high-voltage cable TMR high-frequency partial discharge pulse signal monitoring device architecture diagram. Figure 2 As shown, the device of this embodiment includes: a TMR high-frequency current sensor, a field signal receiver and a back-end processor. The TMR high-frequency current sensor is installed at the top of the cross-connection box and between the cross-connection lines to sense the pulse current signal of the local release at the grounding wire of the shielding layer of the high-voltage cable joint; the field signal receiver is used to receive the local discharge pulse current signal transmitted by the TMR current sensor through the coaxial cable and transmit the data to the back-end processor through the WAPI private design protocol; the back-end processor mainly includes a signal processing module and a server module. The signal processing module conditions the local discharge pulse current signal, performs AD conversion, denoises, and generates a partial discharge spectrum, etc. The server reads the spectrum obtained by denoising the signal processing module, and identifies the features through a multi-scale network to obtain the corresponding partial discharge type.
[0039] Preferably, the number of the TMR high-frequency current sensors is 6, and the first 3 are respectively installed at the joints of the first and second sections of the A phase, the third and first sections of the B phase, and the second and third sections of the C phase of the three-phase crossed high-voltage cable; the last 3 are respectively installed at the joints of the second and third sections of the A phase, the first and second sections of the B phase, and the third and first sections of the C phase.
[0040] Preferably, the WAPI private design protocol mainly sets keys and verification mechanisms for the field signal receiver and the back-end processor through the WAPI authentication server, and at the same time uses encryption algorithms and security authentication protocols to prevent unauthorized external nodes from intercepting and tampering with the high-frequency pulse signal.
[0041] Preferably, the signal processing module mainly includes a receiving unit, a signal conditioning unit, an AD conversion unit, a denoising unit, and a spectrum generating unit.
[0042] The signal receiving unit is used to send ACK key verification and data encryption request to the WAPI authentication server, and receive the WAPI frame simulation current pulse data transmitted by the field signal receiver and perform protocol decryption, and finally transmit it to the signal conditioning unit.
[0043] The wireless network is deployed using wireless controller AC and FIT AP. External field modules and backend processing access network access points, and the LAN port addresses are configured with the same network segment IP address to ensure intercommunication. AP covers a diameter of 20-30 meters in an open environment, and the number and distribution of use are determined according to the environment and user scale.
[0044] Create a WLAN-BSS1 interface, which belongs to the default VLAN 1. Create a security profile named "gaoyadianlan", configure it to use the WAPI security policy, and set the wireless password. Create a service set named "gaoyadianlan", and specify the SSID as "gaoyadianlan".
[0045] The WAPI certificate, confidentiality template and service set can be set individually according to the actual situation.
[0046] Figure 3 : is a module diagram of the signal conditioning circuit of the present invention. Figure 3 As shown, an amplifier circuit, an active filter circuit, and an addition circuit are built as a signal conditioning unit, which are used for impedance transformation and charge conversion, that is, converting the high-impedance charge output of the sensor into a low-impedance amplitude output, and signal amplification, that is, amplifying the weak electrical signal output by the sensor, and preliminarily filtering out high-frequency interference to improve the signal-to-noise ratio. It mainly includes a charge conversion circuit, an active filter circuit, and an output amplifier circuit. The high-impedance charge output of the sensor first passes through the charge conversion circuit to become a low-impedance voltage output, and the signal is preliminarily amplified. The active filter circuit is designed as a second-order active low-pass filter circuit according to the detection needs, with a cutoff frequency of 120kHz to filter out high-frequency interference; the output amplifier circuit further amplifies the signal.
[0047] This embodiment adopts the AD7606 chip as the A / D conversion unit. The AD conversion unit is used to convert the voltage output signal of the sensor after signal conditioning into a digital signal, complete automatic acquisition and send it to the denoising unit for analysis and processing. The AD unit sampling rate selected in this embodiment is adjustable from 500kHz to 1MHz, the number of sampling points is 65536, and the acquisition time is adjustable from 60 to 120ms according to the sampling frequency. It can complete the acquisition of multiple power frequency cycles, which meets the monitoring requirements of partial discharge.
[0048] The signal denoising unit is used to denoise the digital signal, and the denoising method adopted is an improved wavelet threshold denoising method.
[0049] Figure 4 : is a flow chart of the improved wavelet threshold denoising method of the present invention. Figure 4 As shown, the wavelet threshold denoising method of this embodiment improves the traditional wavelet threshold function, making the threshold function adaptive. By adjusting the variable parameters to adapt to the signal filtering of different characteristics, the filtering effect of wavelet threshold denoising in actual conditions is greatly improved. At the same time, the signal-to-noise ratio is improved while ensuring that the target signal is not distorted.
[0050] The improved wavelet threshold function is:
[0051]
[0052] in, represents the kth wavelet coefficient at the jth decomposition scale in the wavelet transform, o is the universal threshold of the wavelet coefficient, a is an adjustable parameter, is the optimized coefficient.
[0053] When the wavelet coefficient When it is less than the threshold, it is considered to be generated by noise and set to zero. Otherwise, it is considered that the coefficient comes from the useful signal and threshold processing is performed on it. The value of a is a non-negative number. When a=0, At this time, the improved threshold function can achieve a soft threshold effect. When a approaches infinity, the threshold function is equivalent to the hard threshold function. Therefore, by adjusting the parameter a, the performance of wavelet threshold denoising can be affected. In order to denoise the signal more accurately, simply and effectively, this embodiment adopts an improved threshold calculation method based on the general threshold formula. Because the wavelet coefficients of the noise are mainly concentrated in the first layer detail coefficients and decrease with the increase of scale, the threshold should also change when the scale changes. Since the traditional general threshold does not conform to the law of noise with wavelet transform, it affects the denoising performance of the signal. The improved threshold o formula is:
[0054]
[0055] Where j is the decomposition scale, N j is the number of high-frequency coefficients in the jth layer.
[0056] It can be seen from the formula that the setting of the variable parameter a value requires analysis of the characteristics of the actual signal and the interference noise, and a rough initial value is given based on past experience. The selection of the a value will have a great impact on the noise reduction effect of the signal. Therefore, this embodiment judges whether the parameter a in the current improved threshold function is applicable by comparing the intensity of the noise content in the filtered signal. If the noise intensity contained in the filtered signal increases, the parameter a needs to be adaptively adjusted, otherwise it remains unchanged. The a value is corrected through continuous feedback to improve the signal-to-noise ratio.
[0057] Preferably, the server includes a feature extraction unit, which is used to extract feature information from artificially produced or historical partial discharge phase distribution maps, and obtain the discharge type corresponding to the historical partial discharge phase distribution map; then use the historical feature data and the corresponding discharge type to train a multi-scale neural network to obtain a discharge type recognition model; a discharge type recognition unit, which is used to input the historical PRPD map into the discharge type recognition model to obtain the discharge type corresponding to the PRPD map.
[0058] Preferably, the local discharge types include corona discharge, surface discharge, internal discharge, and arc discharge. The diagnostic identification database is built with the PRPD spectrum as the data set. In order to make the diagnostic results universal, for the same defect, the test data of 2 cables are used as the test set data, and the test data of another cable is used as the training set data. The test set and the training set are divided at a ratio of 2:1.
[0059] Figure 5 It is a multi-scale neural network model diagram of the present invention. Figure 5 As shown, preferably, a multi-scale neural network model is built, including three branch models of different scales, which can effectively extract feature abstractions of different scales, fuse these features through full connection, and finally output them using the Softmax function. The model is mainly divided into two parts: feature extraction and feature connection; the first is the feature extraction part. With the increase of network layers, the number of feature maps increases exponentially, and the size of feature maps also decreases. The second is the feature connection part. Through feature extraction at different layers, features of different scales are connected to the cascade layer; through full connection, feature fusion is realized in the output layer, and finally the defect failure results are comprehensively output.
[0060] In traditional machine learning theory, the complexity of the model is proportional to the amount of training data. If the model is complex and the training data is insufficient, overfitting will occur; on the contrary, if the model is simple and there is more training data, the performance is easily underestimated. For the partial discharge diagnosis problem of this embodiment, the quality of feature extraction directly affects the final classification effect. Therefore, it is necessary to analyze the values of feature dimensions and the number of neuron outputs, and the optimal values of feature dimensions and the number of neuron outputs can be obtained through intersection verification.
[0061] Therefore, the network modeling is carried out by comprehensively considering the classification accuracy and computational efficiency. In order to obtain higher classification accuracy and less time consumption, the number of neurons in the output layer is set to 512 and the feature dimension is 8. The test set is evaluated based on the established multi-scale convolutional neural network, and the overall prediction accuracy is 94.5%. This method can accurately diagnose high-voltage cable terminal faults, and provides a practical identification method for realizing intelligent and accurate diagnosis of power distribution network cable faults.
[0062] The overall process is as follows:
[0063] 1) The original PRPD (phase resolved partial discharge, PRPD) data was preprocessed and the data set was divided into a training set and a test set with a ratio of 2:1.
[0064] 2) Build a multi-scale neural network model, seek the optimal network parameters through cross-validation, and complete network training.
[0065] 3) Import the collected PRPD samples into the trained network to obtain the recognition results.
[0066] The field signal receiver of the device of this embodiment is used to receive the partial discharge pulse current signal transmitted by the TMR current sensor through the coaxial cable and transmit the data to the back-end processor through the WAPI private design protocol; the back-end processor mainly includes a signal processing module and a server module. The signal processing module conditions the partial discharge pulse current signal, performs AD conversion, denoises, and generates a partial discharge map, etc. The server reads the map obtained by denoising the signal processing module, identifies the features through a multi-scale network, and obtains the corresponding partial discharge type. The high-voltage cable partial discharge monitoring device of this embodiment can realize the spatiotemporal synchronous detection of high-frequency partial discharge signals of high-voltage cables, and is highly portable and has the advantages of integration, low power consumption, strong safety, and accurate identification.
[0067] Those skilled in the art will appreciate that all or part of the steps to implement the above embodiments may be accomplished by hardware, or may be accomplished by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, wherein the storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.
[0068] The above is only a preferred embodiment of the present invention. It should be pointed out that ordinary technicians in this technical field can make several improvements and supplements without departing from the principle of the present invention. These improvements and supplements should also be regarded as the scope of protection of the present invention.
Claims
1. A method for monitoring high-voltage cable TMR high-frequency partial discharge pulse signals, characterized in that The following steps are involved: S1, using TMR high-frequency current sensor to sense the local discharge pulse current signal released at the grounding wire of the shield layer of the high-voltage cable joint; S2, the signal processing module conditions and AD-converts the partial discharge pulse current signal, performs denoising processing using an improved wavelet threshold denoising method with an adaptive threshold function, and generates a partial discharge pulse phase distribution spectrum; S3, the server module inputs the partial discharge pulse phase distribution map into the trained multi-scale neural network model, identifies the partial discharge pulse current signal characteristics, and obtains the corresponding partial discharge type; The threshold function in step S2 has an adaptive improved wavelet threshold denoising method: Improved wavelet threshold function in, represents the kth wavelet coefficient at the jth decomposition scale in the wavelet transform, o is the universal threshold of the wavelet coefficient, a is an adjustable parameter, is the optimized wavelet coefficient; Improved universal threshold of wavelet coefficients Among them, j is the decomposition scale, N j is the number of high-frequency coefficients in the jth layer; By comparing the intensity of the noise content in the filtered partial discharge pulse current signal, it is determined whether the adjustable parameter a of the current improved wavelet threshold function is applicable. If the noise intensity increases, the adjustable parameter a is adaptively adjusted, otherwise it remains unchanged. Step S3: The multi-scale neural network model diagram includes several branch models of different scales, which are divided into two parts: feature extraction and feature connection. First, in the feature extraction part, as the network layers increase, the number of feature maps increases exponentially, and the size of feature maps also decreases accordingly. Secondly, in the feature connection part, through feature extraction of different layers, features of different scales are connected to the cascade layer. Finally, through full connection, feature fusion is realized in the output layer, and the recognition result is comprehensively output using the Softmax function.
2. A method for monitoring high-voltage cable TMR high-frequency partial discharge pulse signals according to claim 1, characterized in that Step S3: The local discharge types include four types: corona discharge, surface discharge, internal discharge, and arc discharge.
3. A high-voltage cable TMR high-frequency partial discharge pulse signal monitoring device, characterized in that: It comprises a TMR high-frequency current sensor, a field signal receiver and a back-end processor; the field signal receiver is used to receive the local discharge pulse current signal transmitted by the TMR current sensor through a coaxial cable and transmit it to the back-end processor; the back-end processor comprises a signal processing module and a server module; the monitoring device is used to execute a high-voltage cable TMR high-frequency local discharge pulse signal monitoring method as described in any one of claims 1 to 2.
4. A high-voltage cable TMR high-frequency partial discharge pulse signal monitoring device according to claim 3, characterized in that: It includes several TMR high-frequency current sensors, some of which are installed at the joints of the first and second sections of phase A, the third and first sections of phase B, and the second and third sections of phase C of the three-phase crossed high-voltage cable of the power grid, and the other part of the TMR high-frequency current sensors are installed at the joints of the second and third sections of phase A, the first and second sections of phase B, and the third and first sections of phase C of the three-phase crossed high-voltage cable of the power grid.
5. A high-voltage cable TMR high-frequency partial discharge pulse signal monitoring device according to claim 3, characterized in that: It also includes a WAPI authentication server; the field signal receiver transmits the local discharge pulse current signal data to the back-end processor through the WAPI private design protocol, and the WAPI authentication server is used to set keys and verification mechanisms for the field signal receiver and the back-end processor, and prevent unauthorized external nodes from intercepting and tampering with the local discharge pulse current signal data through encryption algorithms and security authentication protocols.
6. A high-voltage cable TMR high-frequency partial discharge pulse signal monitoring device according to claim 3, characterized in that: The signal processing module includes a signal conditioning unit, an AD conversion unit, a denoising unit and a spectrum generating unit; the signal conditioning unit includes a charge conversion circuit, an active filtering circuit and an output amplifier circuit, the AD conversion unit is used to convert the output voltage of the partial discharge pulse current signal after signal conditioning into a digital signal and complete automatic collection, and the denoising unit is used to denoise the digital signal.
7. A high-voltage cable TMR high-frequency partial discharge pulse signal monitoring device according to claim 3, characterized in that: The server module includes a feature extraction unit and a discharge type identification unit.
