Neural Network‐Based Adaptive Dynamic Surface Course Tracking Control of an Unmanned Surface Vehicle With Signal Input Quantization

ABSTRACT This paper investigates the adaptive neural network‐ controlled course tracking of an unmanned surface vehicle (USV) with quantization of signal input. As a first step, the characteristics of the ship's rudder servo system are fully considered and combined with the mathematical representation of the ship's heading control system. This is done to develop a nonlinear third‐order response model. The Radial Basis Function (RBF) neural network is constructed to estimate and approximate the unknown functions within a mathematical model of the system, and nonlinear damping terms are employed to counteract external disturbances. Subsequently, a design method for a neural network adaptive quantization controller is proposed. This controller can enable real‐time learning and adjustment to address performance degradation caused by signal quantization errors. Based on the Lyapunov theorem, the designed controller has been validated for its dynamic response capability and system stability, ensuring long‐term reliable and stable operation. In addition, semiglobally, uniformly bound signals are used in closed‐loop systems. Tracking errors are lowered through parameter tuning to trim levels arbitrarily. As a final result, simulation results confirmed the effectiveness and feasibility of the RBF neural network‐based adaptive quantification control method for USVs.