Model Reference Adaptive Neural Network Control For Ship Steering Autopilot With Uncertain Nonlinear Systems

This thesis has systematically researched neural network model reference adaptive control and its application to ship's course steering control.In this thesis ,aimed at ship linear mathematical motion model ,the model reference adaptive autopilot is first designed based on the analysis of the model reference adaptive control theory.By combining model reference adaptive control and neural network control ,a novel model reference adaptive neural network control scheme is proposed for the ship steering's nonlinear model.In the procedure ,the RBF neural networks are used to approximate nonlinear systems;The weights can be achieved by use of Lyapunove approach.To compensate the approximation error and attenuate the external disturbance,a robust technology is introduced to the proposed scheme.Asymptotic stability of the closed loop system is established in the Lyapunov theory,and the tracking errors convergence to a neighborhood of zero.At last ,the scheme is appilied to ship's course steering control and the simulation curves show that the desired results are attained.The model reference adaptive neural network control autopilot is simulated under different conditions,such as extern disturbances ,loadings,velocities,etc. The simulation results show the effectiveness and the salient performance of the proposed control scheme.And the comparion between the model reference adaptive neural network scheme and conventional model reference adaptive scheme demonstrates that the performance of the former is better than that of the latter.To verify the performance of the autopilot design in this thesis,ship's mathematical model is researched ,in consideration of the forces and moments and moments from the environmental disturbances,such as winds,waves,currents,etc.By using Matlab's Simulink toolbox,simulation experiments are carried out on ship steering autopilot.