A partial discharge identification method based on EFPI optical fiber sensor and Attention-CLDNN algorithm

By using the method of combining EFPI fiber sensor array and Attention-CLDNN algorithm in high-voltage generators, the problem of local discharge identification in the prior art is solved, and more efficient and accurate local discharge type identification is achieved.

CN119738671APending Publication Date: 2025-04-01YUNNAN ELECTRIC POWER TESTING & RES INST (GRP) CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411820161.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-11
Publication Date
2025-04-01

AI Technical Summary

Technical Problem

The existing local discharge fault identification technology of high-voltage generators is susceptible to electromagnetic interference and environmental noise, and the neural network-based identification algorithm lacks recognition accuracy and stability during complex signal processing.

Method used

The local discharge recognition method based on EFPI fiber sensor and Attention-CLDNN algorithm is adopted to collect local discharge ultrasonic signals through EFPI fiber sensor arrays with different diaphragm thicknesses, and signal processing and identification is performed using the Attention-CLDNN algorithm introducing a bidirectional LSTM layer and the CNN network introducing residual connections.

Benefits of technology

It improves the accuracy and recognition efficiency of feature extraction of local discharge signals, enhances the recognition stability and accuracy in complex industrial environments, and is suitable for large-scale real-time monitoring systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119738671A_ABST
    Figure CN119738671A_ABST
Patent Text Reader

Abstract

The present invention provides a local discharge identification method based on an EFPI optical fiber sensor and an Attention-CLDNN algorithm, and the method comprises the following steps: S1: using an EFPI optical fiber sensor array with different diaphragm thicknesses to simultaneously collect local discharge ultrasonic signals inside a high-voltage generator; S2: preprocessing the collected different local discharge ultrasonic signals, and processing the local discharge ultrasonic signals obtained by the preprocessing through an Attention-CLDNN algorithm introducing a bidirectional LSTM layer to obtain multiple groups of one-dimensional matrices; S3: constructing two-dimensional matrix data according to the obtained multiple groups of one-dimensional matrix data, inputting the two-dimensional matrix data into a CNN network introducing a residual connection for identification, and outputting a local discharge type identification result.
Need to check novelty before this filing date? Find Prior Art