Abstract: In this paper, for a class of ship heading control systems with time-varying parameters and time-delay nature, a neural network adaptive controller based on backstepping technology is proposed. Firstly, the high fidelity 3 degrees of freedom MMG（Maneuvering Modeling Group） model is transformed into a non-affine nonlinear system with time-varying parameters and time-delay. Afterwards, by combining with the DSC（Dynamic Surface Control） and RBF（Radial Basis Function） neural networks, the adaptive controller is carried out based on backstepping technology. Finally, the effectiveness of the neural network adaptive controller is verified by tracking target signal. Besides, the controller can effectively solve the problem of complexity explosion and has simple structure convenient for implementation. Via stability analysis, the results show that all signals in closed-loop are uniformly ultimately bounded.