The Intelligent Control Of Ship Motion Based On Dynamic Neuro-Fuzzy Model

Shipping movement control in water surface is catching more and more attention in these years.With the features of big inertia,large time-delay and nonlinear,the model parameter perturbation caused by ship speed,storm-flow interference, measurement error and so on,the complex uncertainty problems have been existing in this field a long time.All these make the ship course-keeping control problem considerable difficulty.So far it is also the main reason for the discrepancy of traditional design of intelligent control algorithms based on linear control theory and the actual results researchers expected.Therefore,in order to resolve the uncertainties and the realization of the dynamic control of nonlinear systems,this paper presents a method for the ship course intelligent control that based on dynamic neuro-fuzzy model.This article consider the impact for the ship tracking control,which coursed by the uncertainties of the changing speed.By combining RBF neural network and T-S fuzzy model,setting up a dynamic neuro-fuzzy model,network parameters and structure of the rules layer can regulate themselves in learning process.DNFM fully approximation the inverse dynamics of nonlinear ship model Norrbin,whose modeling parameters time-varying with the change speed.An adaptive controller by good trained DNFM in parallel with the PID controller are used to course tracking control.Taking the output of PID controller as independent variable of the adaptive law,the DNFM weight was adjusted further online and ship course intelligent controller was designed for uncertainty ship motion model.In this paper,the simulation takes the COSCO 5446TEU large container ship as an example.By setting two different changes of speed of the ship,the good dynamic neuro-fuzzy model can be successfully established to approach the inverse dynamics ship model.Then through design the intelligent control that based on DNFM,achieved quickly and efficiently track the ship's course.Overcome the impact of modeling parameters uncertainty.