Time-varying model identified based coupled fault diagnosis for high speed trains

Coupled fault of the high speed train information control system (HSTCS) can appear, when operational conditions change abnormally or operators do not react properly or timely. Typically multiple fault ambiguity groups, including motor warnings with over-temperature, over-current, bar broken fault and air gap eccenticity fault, are highly related to speed level and traction/barking forces regulation. However, it is difficult to apply single fault diagnosis based methods to model the relations between fault and speed or alarm prioritization. In this paper, a T-S time-varying model identified based diagnosis scheme is developed for the HSTCS. Firstly, a multivariate detection index is proposed to identify the thresholds off-line and then coupled fault time-varying model is built. Then, the optimal fuzzy model structure and fault characteristics set are established using the clustering algorithm. Then, a fuzzy weighted least square algorithm with parameters convergence is proposed to estimate the coupled fault. Meanwhile, fault isolation techniques with stability analysis are used to provide a clear alarm priority for different fault modes. Finally, through a coupled fault diagnosis experiments using real data of CRH5G, the effectiveness of the proposed algorithm is verified. © 2019, Editorial Office of Control and Decision. All right reserved.