On-line Modeling of AUV’s Maneuvering Motion in Diving Plane Based on SVM

Support vector machines have been validated as an efficient artificial intelligent algorithm for solving classification and regression problems on account of their simple model structure, good generalization and global optimal solution. However, their on-line performances are not yet widely used. A simple on-line algorithm named incremental least square support vector machines is put forward by analyzing the kernel function matrix and is employed to identify the maneuvering parameters of autonomous underwater vehicles in diving plane. Zigzaglike maneuver with 5o fin angle is simulated based on the mathematical model, and the viscous hydrodynamic coefficients are then on-line identified by using the proposed method. The comparison between the identified parameters and the standard derivatives demonstrates the validation of incremental least square support vector machines. This study is useful for maneuvering study and model-based control for underwater vehicles.