Multi-Thruster Cooperative Fault Diagnosis Based on Dynamic Analysis of Similarity

Fault diagnosis plays an important role in ensuring the safety of underwater vehicles. As one of the most crucial components of an underwater vehicle, fault diagnosis of the propulsion system is of great importance. Most existing methods rely on mathematical models or fault sample data, which bring great difficulties to the realization of fault diagnosis. Some characteristics of the underwater vehicle itself can be used for fault diagnosis to reduce the difficulty of implementation. In this paper, a fault diagnosis method based on dynamic analyzing the response characteristics similarity of multi-thrusters is developed. First, the fault feature extraction is achieved by online measurement of the time series distance between the thruster current signals. Then, based on the Bayesian decision theory, the abnormal changes of similarity probability distribution are enhanced. Finally, through the collaborative analysis of the correlation coefficients between the control variables and current signals of each thruster in the same group, the fault location is realized.