Composite learning based adaptive fuzzy control of unmanned surface vehicles with stochastic disturbances

In this paper, we investigate the path tracking control problem for unmanned surface vehicles, focusing on the challenges posed by uncertain dynamics characteristics and external stochastic disturbances. We propose an adaptive control strategy that leverages the approximation capabilities of fuzzy neural networks to tackle the challenges of controller design arising from uncertain functions. We develop a novel system performance prediction model to evaluate the approximation ability of neural networks, even in the absence of critical system information, such as the inertia matrix. Based on this model, we design a composite adaptive update strategy to enhance system control performance. We also conduct a stochastic stability analysis to ensure that the system signals remain probabilistically bounded and that the tracking error converges to a small region around the origin. Finally, we present simulation and experimental results demonstrating the effectiveness of our proposed method.