Study On Soft Computing Based Intelligent Control And Its Application To Ship Steering

In this thesis, soft computing based control algorithms including genetic algorithms (GA), fuzzy control, neural networks (NN) and their different combinations are discussed systematically. About genetic algorithms, its basic principle, operation, schema theorem, and the realization in the computer are discussed firstly. Considering the shortcomings in the genetic algorithms, some improvements are proposed In order to elucidate its strong optimizing ability, genetic algorithms are used to optimize the PID parameters of ship autopilot. The comparison between the different simulation curves shows the validity of genetic algorithms in looking for the best parameters. The principle, design, the advantages and the shortcomings, as well as the T-S model of fuzzy logic control are discussed. In order to utilize the advantages of the genetic algorithms and overcome the shortcomings of the fuzzy logic control, GA-FUZZY control algorithm is proposed and successfully implemented on the fuzzy rules optimization both in the control of the ship steering and the double inverted pendulum. In the neural networks control, the structure and the working principle of the cerebella model articulation controller (CMAC) are discussed firstly, and then based on the mutual supplements and the similarities between FCMAC and fuzzy logic,fuzzified cerebella model articulation controller (FCMAC) is proposed. The learning control system based FCMAC are introduced in detail. Through the example of FCMAC used in the swinging up a pendulum control, the excellent control effects are demonstrated. As an improvement measure to FCMAC, GA-FCMAC control algorithms is proposed and implemented on the inverted pendulum control successfully. Finally, soft computing based control algorithms are applied in the ship steering. The ship model employs the complicated nonlinear ship model with wind, wave, currents disturbances in order to simulate the practical state of ship motion. Through simulation on two actual ship model controlled by FCMAC and GA-FCMAC separately, satisfactory simulation results have been given, which are beneficial to improve the control performances and energy saving of ships. The simulation results illustrate that the combinations of two or more soft computing based intelligent control algorithms can preserve their merits and compensate their defects, bring about better control effects.