The Intelligent Control Of Unmanned Ship Trajectory Based On Disturbance Observer And MLP

This paper is based on the research of underactuated surface ships,considering that unmanned ship system has stronger uncertainty,more susceptible to external interference and higher technical requirements,combined with the research progress of ship trajectory tracking control,the unmanned environment is introduced,and the original sliding mode control method and RBF neural network algorithm are further improved to adapt to unmanned navigation environment.Specifically completed the following research work:(1)Establishment of mathematical model of unmanned ship.Considering the difference between unmanned ship and traditional ship,a mathematical model of ship motion with non-diagonal terms is introduced.At the same time,coordinate transformation method is used to eliminate the effect of bow moment on lateral velocity in ship model with non-diagonal term,which lays a foundation for the design of controller.(2)Design of disturbance observer.Considering the characteristics that unmanned ships are affected by external disturbances and need to deal with the control autonomously,a nonlinear disturbance observer is designed to estimate and cancel out the external disturbances through the disturbance observer,and then the influence of external disturbances on the controller design is solved.(3)Integral sliding mode controller is designed.Based on the idea of stabilizing the trajectory tracking error by stabilizing the velocity error,the first and second order sliding mode surfaces are designed in the dynamic loop,from which the lateral thrust sliding mode control law and the longitudinal bow moment sliding mode control law are obtained.Considering the possible chattering problem,the virtual control quantity in the form of hyperbolic tangent function is adopted to reduce the chattering.(4)RBF neural network and MLP are introduced.Based on the uncertainty of model parameters caused by extermal disturbances,RBF neural network is introduced to approximate the uncertainties by its universal approximation characteristics,so as to counteract the disturbances.At the same time,in order to reduce the computational load of the controller,MLP method is used to replace the learning of all values with the learning of single parameter,so that the controller is more in line with the practical application of engineering.(5)The simulation results of the above control methods are in accordance with the demonstration results of the theoretical part.Under the influence of external disturbances,the ship can track the given trajectory quickly and accurately,which verifies the feasibility and effectiveness of the control method.