모터 전류 신호를 이용한 전동차 출입문 고장진단 연구

초록

Purpose: Passenger train door systems can undergo failure during operation, causing incon- veniences such as service delays, breakdowns or even resulting in the human casualties. This study proposes a diagnostic process to predict such failure conditions so they can be avoided in advance. Methods: A door test rig is prepared, in which the motor signals are acquired for different fault modes during the operation test. Various time domain features are extracted and selected using the Fisher Discriminant Ratio and Pearson correlation. Classifications are then performed by the k-Nearest Neighbor (k-NN) method for the door-open and door-close operations, respectively. Results: Overall, the fault modes are identified quite well except for the artificial bearing wear. The result is worse in thedoor-close operation because the high peak current applied near the end to ensure the door closing may have affected the classification accuracy. Conclusion: The proposed fault diagnosis method is easy to implement for the passenger train door system. Once the diagnosis system is trained via the seeded fault modes, it can be applied with accuracy to the fault diagnostics of train door systems, and be used for the efficient maintenance.

키워드