RBF-Learning-Based Many-Objective Metaheuristic for Robust and Optimal Controller Design in Fixed-Structure Heading Autopilot

This paper presents an innovative many-objective metaheuristic (MnMH) algorithm designed to tackle the challenges of robust and optimal controller design for fixed-structure heading autopilots. The proposed approach leverages the radial basis function (RBF)-learning operator during the reproduction phase of the MnMH to generate high-quality solutions. A key feature of the method is its generation of a target Pareto front, in which a z-surrogate model makes predictions to guide design solutions toward achieving optimal performance. The effectiveness of the new algorithm is validated through both fixed-structure heading autopilot controller design problems and standard benchmark optimization problems. Results consistently show that the proposed algorithm outperforms several existing MnMHs across all tested scenarios. This study offers valuable insights into many-objective optimization and demonstrates the algorithm’s potential for enhancing robust controller design in heading autopilot systems.