Research On Composite Adaptive Control Of Ship Dynamic Positioning System

With the rapid development of marine engineering equipment,the dynamic positioning and autonomous navigation of ship have attracted widespread attention.The dynamic positioning system is a closed-loop feedback control system which make use of the thrust produced by the propellers to counteract the external environment force to keep the ship or platform in a predetermined position and trajectory.The rapidity,accuracy,and stability of the ship dynamic positioning have become the focus of research in dynamic positioning control systems under the conditions of unknown external environmental loads and uncertain model parameters.By constructing the adaptive law and combining several different control ideas to design a composite adaptive controller to solve the uncertainties becomes the focus of this paper.Firstly,taking into account of the positioning error caused by the static error,a backstepping integral controller and a backstepping neural network adaptive controller are designed.The former controller adapts the external environmental force through the position error integral term,but due to improper parameter setting,it may cause the integral saturation and the disadvantage that the static error is difficult to eliminate.In this thesis,according to the strong approximation performance of orthogonal neural networks,the external environmental forces are estimated by using the orthogonal neural network compensation term instead of the integral term in the backstepping integral controller,and the control law and the weight adaptive law are derived by the backstepping method.Stability analysis are made to ensure that the control system is uniformly bounded.The simulation results and experimental results show that the designed controller can achieve the accuracy of the ship positioning under unknown environmental disturbances.Taking into account of the parameter uncertainties in the ship model,it may cause controller instability.In order to avoid the influence of model uncertainty,this thesis designs a neural network robust controller with backstepping method.Lyapunov stability proof guarantees that the system is uniformly ultimately bounded.However,due to the oscillation of the control output force caused by the positive and negative alternation of the sign function in the robust term,this thesis designs a robust adaptive controller for the neural network by constructing an adaptive law to estimate the upper bound value of the robust term.Ultimately,the composite controller can achieve the stability and accuracy of dynamic positioning considering the uncertain model parameters.Considering slow speed of the controller estimation under changing sea conditions,a composite adaptive controller based on prediction error and tracking error is proposed.The two kinds of errors are used to design the neural network weight adaptive law,and the parameter error is corrected by using the parameter information contained in the prediction error to achieve rapid convergence of parameters.Finally,the simulation on the ship station keeping under the changing environment shows that the stability,accuracy and rapidity of designed composite controller.