Simultaneous Faults Detection and Location of Thrusters and Sensors for Autonomous Underwater Vehicle

Aiming at the problem of detection and location when faults of thruster and sensor for autonomous underwater vehicle occur simultaneously, a method of quantitative /qualitative hybrid diagnosis is proposed, combining neural networks technology with dynamic trend analysis technology. Firstly, a fault detection observer model is proposed to achieve the joint estimation of quantity of state and fault vector for AUV, and the decoupling problem of fault type when the faults of thruster and sensor occur is solved, and then based on dynamic trend analysis theory, the fast location of fault part is achieved by extraction, identification and aggregation of the real-time trends for AUV controlled variables and state measured values, and matching with the trend characteristic sets of established fault knowledge base. Through simulating the simultaneous faults of thruster and sensor, the pool experiment verification of experimental prototype is made, and the results show that the methods proposed in this paper are effective.