Answers are grounded in this paper and cite their sources.
Wind disturbances and changes in rudder effectiveness can degrade ship path-following performance. This study develops a compensation method that combines an offline-trained multilayer perceptron (MLP) with an online disturbance observer (DOB) while retaining line of sight (LOS) guidance and a proportional-derivative (PD) autopilot. The MLP predicts finite-window motion residuals, and the DOB estimates the equivalent deviations remaining after nominal-model and MLP correction. A bounded dual-channel trimming calculation converts these corrections into heading and rudder-angle compensation. Simulations use a four-degree-of-freedom S175 ship model under normal wind, reduced rudder effectiveness, and background wind speeds exceeding the training range. Across 72 test voyages, mean root mean square cross-track errors are 0.702, 0.880, and 2.217 m, respectively, approximately 11% below DOB-only control. Repeated channel comparisons show mean error reductions of 6.9% and 10.2% relative to MLP-only control in two mismatch conditions. Observer replay tests identify reduced estimation bias together with increased sensitivity of sway estimates to measurement noise. A 27-combination parameter study quantifies the effects of controller settings on tracking accuracy and steering activity. This study can support the use of combined residual correction and dual-channel trimming for wind-disturbed ship path following.
No takes yet. Share an insight, caveat, or question.
Sun et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: