Title: Intelligent fault prediction method for traction transformers based on IGWO-SVM and QPSO-LSTM

Authors: Haigang Zhang; Zizhuo Wang; Haoqiang Zhou; Song Zeng; Ming Yin; Junpeng Xu; Bulai Wang; Jinbai Zou

Addresses: School of Rail Transit of Shanghai Institute of Technology, No. 100 Haiquan Road, Shanghai, 201418, China ' School of Rail Transit of Shanghai Institute of Technology, No. 100 Haiquan Road, Shanghai, 201418, China ' School of Rail Transit of Shanghai Institute of Technology, No. 100 Haiquan Road, Shanghai, 201418, China ' School of Rail Transit of Shanghai Institute of Technology, No. 100 Haiquan Road, Shanghai, 201418, China ' School of Rail Transit of Shanghai Institute of Technology, No. 100 Haiquan Road, Shanghai, 201418, China ' School of Rail Transit of Shanghai Institute of Technology, No. 100 Haiquan Road, Shanghai, 201418, China ' School of Rail Transit of Shanghai Institute of Technology, No. 100 Haiquan Road, Shanghai, 201418, China ' School of Rail Transit of Shanghai Institute of Technology, No. 100 Haiquan Road, Shanghai, 201418, China

Abstract: To address the issue of imprecise fault diagnosis and forecasting in traction transformers, this study introduces a composite algorithm that integrates IGWO-SVM and QPSO-LSTM to realise the fault prediction of traction transformer. In this research, the focus is on utilising the concentration levels of dissolved gases in the oil of traction transformers as training data. A fault diagnosis model is constructed leveraging the capabilities of support vector machine (SVM) technology. To improve the performance of the model, this paper adopts an improved grey wolf optimisation algorithm (IGWO), which adjusts the parameters of SVM, enabling it to diagnose the faults of traction transformer effectively. To predict the faults of the traction transformer, this study introduces a model based on the long short-term memory (LSTM) network, which is further enhanced by the integration of the quantum particle swarm optimisation algorithm (QPSO). The QPSO algorithm is employed to fine-tune the parameters of the LSTM network, thereby enhancing the precision and efficiency of its predictive capabilities. Through experimental comparisons with alternative methodologies, the effectiveness and superiority of the proposed algorithm in fault diagnosis and prediction are verified, and the application value of the proposed algorithm in the field of traction transformer fault prediction is demonstrated.

Keywords: traction transformer; fault prediction; gas concentration; improved grey wolf optimisation algorithm; IGWO; support vector machine; SVM; long short-term memory; LSTM; quantum particle swarm optimisation algorithm; QPSO.

DOI: 10.1504/IJPEC.2024.141868

International Journal of Power and Energy Conversion, 2024 Vol.15 No.4, pp.408 - 424

Received: 14 Dec 2023
Accepted: 07 Apr 2024
Published online: 02 Oct 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article