The Adaptive NN Control Design For Ship’s Linear-Tracking

The appearance of nonlinear geometric theory provides favorable conditions for the research of nonlinear control systems. Backstepping is an important tool in the domain of nonlinear control. Considering the nonlinearity of the ships, the method of backstepping is employed and becomes a hot topic in the domain of ship tracking control.In this paper, neural network controlling algorithm is designed separately for ship motion models with and without disturbances on the basis of backstepping. The method divides the whole controlled system into some subsystems, which the amount of is same with the order of the controlled system. By designing virtual control law for each subsystem, we can get the control law of the input. By employing Radical Basis Function Neural Network (RBFNN), the uncertain functions are approximated gradually, and by employing Nussbaum gain technology, the problems of uncertain control gain in the system are solved. With the increase of the order of the system, the backstepping would make the complexity of calculation increase and this problem could be solved by Dynamic Surface Control (DSC).By analyzing the stability of the ship, this algorithm is proved that it could make the closed system Semi-global Uniformly Ultimately Bounded (SGUUB). The simulation results of Matlab show that the algorithm has the property of robustness.