Model Identification for Autonomous Underwater Vehicles Based on Maximum Likelihood Relaxation Algorithm

In order to obtain a precise mathematical model of an autonomous underwater vehicle (AUV), model identification based on maximum likelihood relaxation algorithm is proposed in this paper. With experiment data of zigzag motion, the hydrodynamic derivatives of the AUV are estimated, and ideal convergence of maximum likelihood relaxation algorithm can be obtained from the contrast with normal approach. The simulation system based on these parameters is established to verify the validity. The results show the model is credible, which is very useful for the research of maneuverability and adaptive control of AUVs.