CN-121978484-A - Partial discharge signal identification method, device, apparatus, medium and program product

CN121978484ACN 121978484 ACN121978484 ACN 121978484ACN-121978484-A

Abstract

The application relates to a partial discharge signal identification method, a device, equipment, a medium and a program product. The method comprises the steps of obtaining original partial discharge pulse data of equipment to be detected, inputting the original partial discharge pulse data into a pre-trained partial discharge signal enhancement model to obtain enhanced partial discharge pulse data, matching the enhanced partial discharge pulse data with a standard pulse signal to obtain candidate partial discharge pulse data, obtaining response characteristic information of the equipment to be detected, inputting the candidate partial discharge pulse data and the response characteristic information into a partial discharge identification model, respectively executing characteristic extraction to obtain a partial discharge pulse characteristic to be detected and response distribution characteristics, and obtaining a target response identification result of the equipment to be detected based on the partial discharge pulse characteristic to be detected and the response distribution characteristics. By adopting the method, multidimensional features such as a topological structure, an operation state and the like of the equipment can be fused, the identification accuracy is high, and the anti-interference capability is strong.

Inventors

DENG HAO
LIU GUOWEI
LI ZHE
SUN CHUANG
HE WEI
Dong Dingyi
Zhu Caile
HUANG DONGYI

Assignees

深圳供电局有限公司

Dates

Publication Date: 20260505
Application Date: 20260313

Claims (10)

1. A method for identifying partial discharge signals, the method comprising: Acquiring original partial discharge pulse data of equipment to be detected; Inputting the original partial discharge pulse data into a pre-trained partial discharge signal enhancement model to obtain enhanced partial discharge pulse data; matching the enhanced partial discharge pulse data with a standard pulse signal to obtain candidate partial discharge pulse data; acquiring response characteristic information of the equipment to be detected; And inputting the candidate partial discharge pulse data and the response characteristic information into a partial discharge recognition model, respectively executing characteristic extraction to obtain a to-be-detected partial discharge pulse characteristic and a response distribution characteristic, and obtaining a target response recognition result of the to-be-detected equipment based on the to-be-detected partial discharge pulse characteristic and the response distribution characteristic.
2. The method of claim 1, wherein the matching the enhanced partial discharge pulse data with a standard pulse signal to obtain candidate partial discharge pulse data comprises: performing time domain alignment matching on the original partial discharge pulse data and the standard pulse signal to obtain a target signal segment of the original partial discharge pulse data in a candidate sampling point set; comparing each pulse on the target signal segment with a corresponding pulse on the standard pulse signal to determine a background pulse and a partial discharge pulse in the target signal segment; Constructing discharge type time sequence distribution and pulse time mark distribution of the equipment to be detected according to the standard pulse signal, the background pulse and the partial discharge pulse; and performing time domain feature mapping on the discharge type time sequence distribution and the pulse time mark distribution according to a standard pulse time stamp of the pulse in the standard pulse signal to obtain the candidate partial discharge pulse data.
3. The method of claim 2, wherein the partial discharge pulses comprise at least one of free discharge characteristic pulses, intermittent interrupt characteristic pulses, and phase distortion characteristic pulses, the discharge type timing distribution comprises a standard power frequency pulse timing distribution and a discharge characteristic topology, the pulse time stamp distribution comprises a set of periodic phase coordinates and a standard pulse time stamp distribution, and the constructing the discharge type timing distribution and pulse time stamp distribution of the device to be detected from the standard pulse signal, the background pulse, and the partial discharge pulses comprises: Pulse time sequence integration is carried out on the discharge source spectrum identification of each pulse in the standard pulse signal corresponding to the target signal section and the free event marking quantity of each free discharge characteristic pulse, so that the standard power frequency pulse time sequence distribution is obtained; Pulse time sequence integration is carried out on the discharge source spectrum identification of the background pulse, the discharge source spectrum identification of the phase distortion characteristic pulse, the discharge source spectrum identification of the free discharge characteristic pulse and the oscillation missing flag bit of each intermittent interruption characteristic pulse, so as to obtain the discharge characteristic topology; Pulse time sequence integration is carried out on the periodic phase coordinates corresponding to each pulse in the discharge characteristic topology and the periodic phase coordinates of each intermittent interruption characteristic pulse to obtain the periodic phase coordinate set, and pulse time sequence integration is carried out on the standard pulse time stamp of each pulse in the discharge characteristic topology and the standard pulse time stamp of each intermittent interruption characteristic pulse to obtain the standard pulse time stamp distribution.
4. The method according to claim 1, wherein the obtaining response characteristic information of the device to be detected includes: Acquiring equipment topology description, dynamic measurement point parameters and infrared thermal image characteristics of the equipment to be detected; performing parameter feature coding on the equipment topology description according to a preset topology coding mapping to obtain a topology feature index of the equipment to be detected; Carrying out thermal image feature extraction on the infrared thermal image features to obtain a temperature field distribution spectrum of the infrared thermal image features, and carrying out multi-source feature fusion on the dynamic measurement point parameters and the temperature field distribution spectrum to obtain a state feature vector of the equipment to be detected; and obtaining the response characteristic information based on the topological characteristic index and the state characteristic vector.
5. The method according to claim 1, wherein the partial discharge recognition model includes a feature extraction module, a feature fusion module, and a state classification module, the state classification module includes a feature aggregation component and a state predictor, the candidate partial discharge pulse data and the response feature information are input into the partial discharge recognition model, feature extraction is performed respectively, to obtain a to-be-detected partial discharge pulse feature and a response distribution feature, and a target response recognition result of the to-be-detected device is obtained based on the to-be-detected partial discharge pulse feature and the response distribution feature, including: Extracting the to-be-detected partial discharge pulse characteristics of the candidate partial discharge pulse data based on the characteristic extraction module, and extracting response distribution characteristics of the response characteristic information; Performing feature cascading on the to-be-detected partial discharge pulse features and the response distribution features based on the feature fusion module, and performing feature space expansion on the features after feature cascading to obtain a multi-source discharge feature field; Based on the characteristic aggregation component, carrying out characteristic space compression on the multi-source discharge characteristic field to obtain one-dimensional to-be-detected partial discharge pulse characteristics of the to-be-detected equipment; And based on the state predictor, carrying out feature refining on the one-dimensional partial discharge pulse feature to be detected, predicting the state attribution confidence level of each piece of response feature information to be detected in the equipment to be detected and the response feature information to be selected in the response feature information set, and determining the response feature information to be selected, the state attribution confidence level of which is in a state judgment threshold value, as a target response recognition result of the equipment to be detected.
6. The method of claim 1, wherein the partial discharge signal enhancement model is trained by: The method comprises the steps of obtaining a high signal-to-noise ratio pure partial discharge pulse sample set, wherein the high signal-to-noise ratio pure partial discharge pulse sample set comprises a plurality of pure partial discharge pulse samples with known discharge types and corresponding intensity levels; acquiring a real background noise sample set, wherein the real background noise sample set comprises background noise samples without partial discharge activity, which are acquired from a complex acoustic environment similar to equipment to be detected; scaling the pure partial discharge pulse samples in the high signal-to-noise ratio pure partial discharge pulse sample set according to different preset amplitude ratios to obtain scaled partial discharge pulse samples with different intensities; mixing each scaled partial discharge pulse sample with a background noise sample randomly selected from the real background noise sample set according to different preset superposition ratios to generate a partial discharge pulse mixed sample under the condition of simulating different signal to noise ratios; taking the partial discharge pulse mixed sample as an input sample, and taking a corresponding non-scaled high signal-to-noise ratio pure partial discharge pulse sample as a target output sample to form the partial discharge signal enhancement training sample set; Training an original partial discharge signal enhancement model by utilizing the partial discharge signal enhancement training sample set, and adjusting model parameters by optimizing a preset loss function, so that the original partial discharge signal enhancement model learns the mapping relation from noise-containing partial discharge pulses to pure partial discharge pulses, thereby effectively enhancing the input low signal-to-noise ratio partial discharge pulses and obtaining the trained partial discharge signal enhancement model.
7. A partial discharge signal recognition apparatus, the apparatus comprising: the first module is used for acquiring original partial discharge pulse data of equipment to be detected; the second module is used for inputting the original partial discharge pulse data into a pre-trained partial discharge signal enhancement model to obtain enhanced partial discharge pulse data; The third module is used for matching the enhanced partial discharge pulse data with the standard pulse signal to obtain candidate partial discharge pulse data; a fourth module, configured to obtain response characteristic information of the device to be detected; and a fifth module, configured to input the candidate partial discharge pulse data and the response characteristic information into a partial discharge recognition model, perform characteristic extraction respectively, obtain a to-be-detected partial discharge pulse characteristic and a response distribution characteristic, and obtain a target response recognition result of the to-be-detected device based on the to-be-detected partial discharge pulse characteristic and the response distribution characteristic.
8. A computer device comprising a memory storing a computer program and a processor implementing the steps of the method of any one of claims 1 to 6 when the computer program is executed by the processor.
9. A computer-readable storage medium, characterized in that a computer program is stored thereon, which, when being executed by a processor, implements the steps of the method of any of claims 1 to 6.
10. A computer program product comprising a computer program which, when executed by a processor, implements the steps of the method of any one of claims 1 to 6.

Description

Partial discharge signal identification method, device, apparatus, medium and program product Technical Field The present application relates to the field of artificial intelligence technology, and in particular, to a method, apparatus, device, medium, and program product for identifying partial discharge signals. Background Partial discharge is an important sign of insulation degradation of power equipment, and accurate detection and identification of partial discharge signals are critical to safe operation of the equipment. The local discharge signals collected on site are often interfered by strong background noise, weak local discharge pulses and complex noise are difficult to effectively separate by traditional signal enhancement methods such as filtering technology, so that the signal to noise ratio is limited to improve, meanwhile, local discharge identification is dependent on manual experience or a simple mode identification algorithm, multidimensional features such as a topological structure and an operation state of equipment are difficult to fuse, the identification accuracy is low, and the anti-interference capability is weak. Disclosure of Invention Based on the above, it is necessary to provide a partial discharge signal identification method, device, equipment, medium and program product, which can integrate multidimensional features such as equipment topological structure and running state, and has high identification accuracy and strong anti-interference capability. In a first aspect, the present application provides a partial discharge signal recognition method. The method comprises the following steps: Acquiring original partial discharge pulse data of equipment to be detected; Inputting the original partial discharge pulse data into a pre-trained partial discharge signal enhancement model to obtain enhanced partial discharge pulse data; matching the enhanced partial discharge pulse data with a standard pulse signal to obtain candidate partial discharge pulse data; acquiring response characteristic information of the equipment to be detected; And inputting the candidate partial discharge pulse data and the response characteristic information into a partial discharge recognition model, respectively executing characteristic extraction to obtain a to-be-detected partial discharge pulse characteristic and a response distribution characteristic, and obtaining a target response recognition result of the to-be-detected equipment based on the to-be-detected partial discharge pulse characteristic and the response distribution characteristic. In some embodiments of the method, the matching the enhanced partial discharge pulse data with a standard pulse signal to obtain candidate partial discharge pulse data includes: performing time domain alignment matching on the original partial discharge pulse data and the standard pulse signal to obtain a target signal segment of the original partial discharge pulse data in a candidate sampling point set; comparing each pulse on the target signal segment with a corresponding pulse on the standard pulse signal to determine a background pulse and a partial discharge pulse in the target signal segment; Constructing discharge type time sequence distribution and pulse time mark distribution of the equipment to be detected according to the standard pulse signal, the background pulse and the partial discharge pulse; and performing time domain feature mapping on the discharge type time sequence distribution and the pulse time mark distribution according to a standard pulse time stamp of the pulse in the standard pulse signal to obtain the candidate partial discharge pulse data. In some embodiments of the method, the partial discharge pulse comprises at least one of a free discharge characteristic pulse, an intermittent interrupt characteristic pulse, and a phase distortion characteristic pulse, the discharge type timing profile comprises a standard power frequency pulse timing profile and a discharge characteristic topology, the pulse time stamp profile comprises a set of periodic phase coordinates and a standard pulse time stamp profile, and the constructing the discharge type timing profile and pulse time stamp profile of the device to be detected from the standard pulse signal, the background pulse, and the partial discharge pulse comprises: Pulse time sequence integration is carried out on the discharge source spectrum identification of each pulse in the standard pulse signal corresponding to the target signal section and the free event marking quantity of each free discharge characteristic pulse, so that the standard power frequency pulse time sequence distribution is obtained; Pulse time sequence integration is carried out on the discharge source spectrum identification of the background pulse, the discharge source spectrum identification of the phase distortion characteristic pulse, the discharge source spectrum identification of the free discharge characteris