Thruster robust fault diagnosis of underwater vehicle

Research on thruster fault diagnosis of underwater vehicles (UVs) is undertaken. UVs work in unknown and very complex ocean environment and an accident occurs on a UV will result in great economical loss. Therefore, it is essential for a UV to have the ability to deal with breaking-out affairs, namely, the ability of autonomous fault diagnosis and fault tolerant control. Sliding-mode observer was brought in UVs´ thruster fault diagnosis. Residuals were obtained by comparing outputs of sliding-mode observer with real sensors. Fault information was distilled by analyzing residuals with fuzzy residual evaluator, which could overcome high rate of misinformation of threshold evaluation method. Results from simulation experiments had shown that the method proposed could successfully distinguish the state change from the state exception of system, and could deduce which thruster was the most possible cause of fault and when the fault was occurred with high precision. The method proposed here provides a solution for the actuator fault diagnosis of UV.