Research on Test-platform and Condition Monitoring Method for AUV

To improve the reliability and intelligence of autonomous underwater vehicle, an AUV test-platform named "Beaver" is developed. The hardware and software system structure are introduced in detail. By analyzing the performance and the fault mechanism of thruster, it establishes the condition monitoring system for thrusters and sensors based on the double closed-loop PID controller, which includes the performance model of thruster based on RBF neural network and forward model of AUV based on improved dynamic recursive Elman neural network, and it probes into the method of combine fault detection. The results of experiment indicate that the "Beaver" can achieve the basic motion control and meet the requirement in test, and the combine fault detection method by parallel connected performance model and forward model can detect the typical fault of thrusters and sensors, which certificates the reliability and the effectiveness of condition monitoring system