During the operation of maglev trains, the actuator fault of the suspension system will affect the normal operation of the trains and may cause serious safety accidents. Therefore, how to make fault diagnosis for the actuator of the suspension system is the premise of train safe operation. In this paper, a fault diagnosis method based on robust observer is used to detect and isolate the actuator fault. First, the fault is decoupled from the system interference through state and output transitions, so that module 1 is related to the fault and interference, while module 2 is only related to the fault. Then, a sliding mode observer is designed for module 2, and the output error is taken as residual error to judge the occurrence of the fault. Eventually, the sliding mode observer is designed for each actuator, and based on the system structure matrix, which actuator position produces the fault is determined. The experimental results show that these schemes mentioned in this paper can successfully realize fault detection and isolation after the actuator broke down, which provides strong support for the safe operation of maglev trains.

maglev trains， suspension system， robust observer， fault diagnosis