Research of system identification method for underwater vehicle based on neural network

Aiming at the characteristic of underwater vehicle which is of big delay and this non-line system is difficult to build the mathematical model, also since the "beaver" underwater vehicle is developed equipped fewer sensors and the thruster has speed feedback, an intelligent model-based method to identify underwater vehicle system is proposed. The motion model of underwater vehicle was built by wavelet neural network, and the performance model of thruster was built based on the improved RBF neural network. Through improving the network structure and algorithm, the model has better approximation capability and faster training speed and provides the reliable data for the following fault diagnosis system of underwater vehicle. It also provides a reference to build models for underwater vehicle motion and thruster. The results of experiment show that the method proposed is effective and feasible.