Application Research On Multiple Model Approach Of Motion Control For AUV

Research on Autonomous Underwater Vehicle motion control system is one of the key research problems on AUV, whose task requirement is very strict such as long work radius, a wide range of regulation capacity of vehicle speed, long work time and the variety of configuring the AUV mission modules. So the system is highly nonlinear, strongly coupled, time-variable and uncertain when it is greatly disturbed by current and wave. The control system must have adequate precision of motion control, strong robust and nonlinear adaptability to adapt above requirements. It brought about austere difficulty to the system design.Multi-model Control is put forward to achieve good performance at various operating points during the motion process. Thus multi-model control is practical for the AUV yaw, attitude, or depth control and track keeping under nonlinear condition.The dissertation mainly worked on multi-model approach and AUV control. Firstly the six DOF kinetic equation, actuator and disturbance model of current and wave are established. Open-loop simulation was done to test the mathematical model. Single model based PID, Sliding Mode, and Fuzzy Controllers are designed to realize the speed, steering and diving control system, from which we found that the control performance far from operation points is not good enough.Then Multi-model Control is studied, in which fuzzy inference rules based switcher is adopted to switch multiple controllers smoothly. Multiple PID, Multiple Sliding Mode and Multi-model based fuzzy controllers are introduced to solve the speed, depth, pitch and yaw control problem respectively.Finally indirect track keeping system with fuzzy bow angle solver is designed to avoid changing steer frequently, and simulations are done in the tracking of straight line, comb line and trapezium path.All in all, multiple model approach covers the wide range variety and uncertainty, when turning the complex problem to several simple systems. And it is self-evident to get good performance while multi-model approach is associated with fuzzy smooth switcher.