Abstract
A disturbance decoupled fault diagnosis strategy is proposed. This disturbance decoupled fault diagnosis is both robust to disturbances and sensitive to sensor faults of magnetic levitation control system. First, a robust controller based on a novel disturbance observer is devised to improve the disturbance attenuation ability, which greatly enhances the robustness of the system. Second, a fault reconstruction technique with adaptive method is presented, along with a strict verification for guaranteeing the robustness of fault. This fault reconstruction technique provides an accurate sensor fault reconstruction. From the results of simulation and experiments conducted on the CMS-04 maglev train, the integrated strategy is robust to model uncertainties of the system and the fault reconstruction algorithm is able to reconstruct the dynamic uncertain faults.