A Design Of Ship Autopilot Based On Improved Genetic Algorithm

This thesis has systematically researched relevant knowledge, such as genetic algorithm, ship’s mathematical model, PID and the applications to ship’s course steering control.To speed up convergence and improve local convergence in genetic algorithm, an adaptive genetic algorithm based on arctangent function has been proposed. The optimal individuals in a population may not be selected into the genetic operation roulette wheel selection, which lead to the best individual missing, to overcome this defect, use the optimal preservation strategy; Using multi-point crossover operator to overcome the inadequate information exchange of single point cross; Nonlinear adaptive crossover probability and mutation probability function are designed to overcome traditional genetic algorithm’s slow convergence speed. They are based on the arctangent function with three parameters of maximal fitness, minimal fitness and average fitness. Simulation results prove that the proposed improved adaptive genetic algorithm possesses faster convergence speed than GA and AGA presented by Srinvas, stronger optimization ability and avoid the premature effectively.The effect of traditional PID controller is not ideal, with bigger overshoot. An adaptive genetic algorithm PID controller based on arctangent function has been designed in this paper. Adjust parameters of PID with the arctangent adaptive genetic algorithm. Compared two PID controllers, the controller designed in this paper and the traditional genetic algorithm PID controller, simulation research shows that, the performance of the IAGA-PID autopilot is greatly improved, and the control system is characterized by little overshoot, short rising time, and excellent robustness.