Keywords
batch standardization;LSTM neural network;online monitoring;fault diagnosis;smart grid;transformer
Abstract
As one necessary equipment in the high‑voltage power system, once the transformer fails, protection devices may refuse to operate and cause the malfunction of power grids. Traditional current transformer fault diagnosis and classification methods firstly extract features from the input operation data, and then use a specific classifier to diagnosis, which lacks adaptive update processing for dynamic input information. In order to further improve the accuracy of traditional recursive neural networks, the process efficiency of long short‑term memory neural networks, this paper proposes a fault diagnosis method based on the LSTM model of batch normalization (BN). This method does not require feature extraction and classifier design steps, where the fault can be classified directly, and can also be updated adaptively. Compared with other fault diagnosis methods, this method has higher diagnostic accuracy and diagnostic performance, which validating its good application value in the field of current transformer fault diagnosis.
DOI
10.19781/j.issn.1673-9140.2023.06.016
First Page
152
Last Page
158
Recommended Citation
CAO, Zhiqiang and CHEN, Jie
(2024)
"Online monitoring and fault diagnosis technology of transformers based on the LSTM with batch normalization,"
Journal of Electric Power Science and Technology: Vol. 38:
Iss.
6, Article 16.
DOI: 10.19781/j.issn.1673-9140.2023.06.016
Available at:
https://jepst.researchcommons.org/journal/vol38/iss6/16