A Robust Data-Driven Course Keeping Control Based on Closed-Loop Gain Shaping Algorithm

Objectives To address the difficulty of accurately modeling ship motion systems caused by strong nonlinearity, time-varying parameters, and environmental disturbances, a robust data-driven heading keeping control method (RDDC) is proposed by integrating the closed-loop gain shaping algorithm (CGSA) with compact-form dynamic linearization model-free adaptive control (CFDL-MFAC). Methods The proposed method employs MFAC to online identify the pseudo partial derivative (PPD) using input-output data, thereby constructing an equivalent dynamic linearized model. The PPD is then embedded into the CGSA framework to design a desired closed-loop transfer function with prescribed dynamic characteristics. After Tustin discretization, an implementable difference-form control law is obtained. Results The bounded-input bounded-output (BIBO) stability of the closed-loop system is theoretically guaranteed. Simulation results under nominal conditions, external disturbances, and model parameter perturbations demonstrate that the proposed method achieves stable and reliable heading keeping. Compared with a representative MFAC-based controller, the proposed method reduces the mean absolute error (MAE), maximum instantaneous error (MIA), and control input variation index (MTV) by 19.5%, 73.6%, and 49.9%, respectively. Conclusions The proposed RDDC method preserves the model-free advantage of data-driven control while enabling structured shaping of closed-loop dynamic performance and enhanced robustness. The results indicate that the method provides an effective solution for ship heading control in complex marine environments and is applicable to autonomous control of nonlinear, time-varying, and uncertain systems.