Robust Adaptive Control For Unmanned Surface Vehicle Considering COLREGS

With the rapid development of intelligent control technology,intelligent navigation control technology has become a important issue in the field of ship control.The problem of intelligent navigation for vehicle can be divided into two parts.The first is path guidance,that is,helping vehicle plan a safe and feasible path to navigate at the current position and target,while avoiding moving or static obstacles or other ship.The other is the path following,which enables the ship’s attitude to navigate along the planned path to ensure navigation safety.For the intelligent navigation scheme in ocean engineering,this thesis proposes a set of appropriate guidance principle considering collision avoidance rules and robust adaptive path following control with disturbance observer.Finally,the effectiveness of the intelligent navigation algorithm is verified by mathematical simulation under ocean environment disturbance.Based on the improved artificial potential field(APF)and dynamic virtual ship(DVS)technology,a novel real time guidance principle considering collision avoidance rules is proposed for tackling intelligent ship path guidance.In the algorithm,the acceptable proximity(AP)is obtained by the relative position and velocity of the own vehicle,obstacles and target.Finally,just like the water seeks its own level,the smooth path is planned from the potential field map formed the AP.Through the above design,the approach is with the capability to avoid obstacles in accordance with the International Regulations for Preventing Collisions at Sea(COLREGS)while pursuing the randomly moving dynamic target.Moreover,if the other vehicle’s operation is improper or the vehicle’s actuator is damaged,the obstacle avoidance behavior is not made in time.This algorithm can make emergency obstacle avoidance operation(EOAO)seasonably according to the real-time vehicle attitude,which effectively reduces the probability of ship collision.Aiming at the path following control,this thesis develops a robust neural path tracking algorithm with disturbance observer(DOB)by means of RBF NNs,minimal learning parameter(MLP),dynamic surface control and backstepping techniques.During navigation,external disturbance is one of the main factors that affect the stability of the ship control system.The proposed control algorithm effectively reduces the randomness owing to environment disturbance by observing,which enhances the controllability of the ship control system.Furthermore,the uncertainties of ship model are compensated and approximated by RBF NNs,and the online updating of the neural network weight parameters is avoided,which can reduce the calculation burden of the closed-loop system.