Abstract:

In order to prevent railway accidents, overall structure of ZPW-2000 track circuit fault diagnosis system was designed, and four major components were clarified, including data pre-processing, data analysis, data service and data application. Firstly, electrical characteristics data of track circuits were compressed by improved revolving gate algorithm. Then, piecewise linear fitting was carried out for analog data in different states of the circuits, and eigenvalues were calculated. Finally, circuit faults were diagnosed by feature extraction method of density clustering, and 9 common ones were identified. The results show that the improved SDT can effectively compress electrical characteristics data of track circuits, eigenvalues of compressed data can be effectively extracted after segmental fitting, and furthermore, density clustering algorithm can be used to generate an effective diagnostic model. Improving fault diagnosis accuracy can help increase maintenance efficiency and capability of signal equipment.

Key words: ZPW-2000, track circuit, fault diagnosis, spinning door transformation(SDT)algorithm, density clustering