Fault diagnosis based on second-order Taylor series dynamic prediction for autonomous underwater vehicle sensor

A sensor fault diagnosis method based on second-order Taylor series dynamic prediction for autonomous underwater vehicle is proposed. Its principle and implementing steps for sensor fault diagnosis are described in detail. The method can improve the prediction accuracy comparing with grey dynamic prediction approach. Thus, the possibility of misjudgment can be decreased. The simulation results demonstrate the effectiveness of it for four typical fault models of autonomous underwater vehicle sensors. In addition, the prediction data can replace the failure data to recover the signal.