Open Access

ARTICLE

A Novel Method for Aging Prediction of Railway Catenary Based on Improved Kalman Filter

by Jie Li^1,3,*, Rongwen Wang², Yongtao Hu^1,3, Jinjun Li¹

1 School of Electrical Engineering and Automation, Henan Institute of Technology, Xinxiang, 453003, China
2 Technology Center, Sichuan Injet Electric Co., Ltd., Deyang, 618000, China
3 Embedded System Research Institute, Xinxiang Engineering Research Center for Intelligent Condition Monitoring of Machinery, Xinxiang, 453003, China

* Corresponding Author: Jie Li. Email:

Structural Durability & Health Monitoring 2024, 18(1), 73-90. https://doi.org/10.32604/sdhm.2023.044023

Received 18 July 2023; Accepted 27 October 2023; Issue published 11 January 2024

Abstract

The aging prediction of railway catenary is of profound significance for ensuring the regular operation of electrified trains. However, in real-world scenarios, accurate predictions are challenging due to various interferences. This paper addresses this challenge by proposing a novel method for predicting the aging of railway catenary based on an improved Kalman filter (KF). The proposed method focuses on modifying the priori state estimate covariance and measurement error covariance of the KF to enhance accuracy in complex environments. By comparing the optimal displacement value with the theoretically calculated value based on the thermal expansion effect of metals, it becomes possible to ascertain the aging status of the catenary. To improve prediction accuracy, a railway catenary aging prediction model is constructed by integrating the Takagi-Sugeno (T-S) fuzzy neural network (FNN) and KF. In this model, an adaptive training method is introduced, allowing the FNN to use fewer fuzzy rules. The inputs of the model include time, temperature, and historical displacement, while the output is the predicted displacement. Furthermore, the KF is enhanced by modifying its prior state estimate covariance and measurement error covariance. These modifications contribute to more accurate predictions. Lastly, a low-power experimental platform based on FPGA is implemented to verify the effectiveness of the proposed method. The test results demonstrate that the proposed method outperforms the compared method, showcasing its superior performance.

---

Graphic Abstract

---

Keywords

---

Supplementary Material

Supplementary Material File

---