Fuzzy Neural Network Based Course Control Of The Rotary Double-Propeller Ship

In severe sea conditions, the ship course control process is often accompanied by frequent rudder angle adjustment and it is difficult to achieve precise course-keeping or tracking. In order to deal with the above problems, this thesis studies ship course control system of rotary electric propulsion ship with doublepropellersunder complex sea conditions based on fuzzy neural network control method. The specific contents are as follows:First, the thesis introducesa rotary electric propulsion ship model with double propellers.Compared with the conventionalstern shaft propulsion ship, the motion characteristics of the electric propulsion ship are more complicated. Specifically, under the action of double propellers, the rotary motion andpropulsive movement of the vessel are highly coupled. And the rotary motion of propellers is equivalent to the rudder movement generated by steering gear.Second, based on the above model of rotary electric propulsion ship with double propellers the T-S fuzzy system is adopted and the course controller is designed. The deviations of ship heading angle and fore roll angular velocity are treated as control inputs and the propeller rotation angle is regarded as a control output. The stability of the obtained closed-loop control system is proved by constructing piecewise smooth Lyapunov function. And the results are in comparison with that of PD controller.Finally,aiming at the ship model parameter uncertainty and fuzzy rule selection problem,this paper introduces the adaptive fuzzy neural networks (AFNN) algorithm. The AFNN control algorithm is adopted to identify the ship model. Thenthe PD controller and robust adaptive control strategy on the basis of adaptive fuzzy neural network compensation are designed. The simulation results demonstratethat the proposed adaptive fuzzy neural network control method has obvious superiority compared with the conventional PD controller.This thesis puts forward a motion model of rotary electric propulsion ship with double propellers, which can meet the maneuverabilityrequierementof small vessels in inland water. AFNN is used for ship model identification and fuzzy control and adaptive fuzzy neural network control algorithm is used for couse controller design. The accuracy and speed of course control is enhanced. In addition, adaptive fuzzy neural networkhas superiority and effectivenessin terms ofthe fuzzy rule selection and control effect.