Efficient Application of Genetic Algorithms for Robust Control System Design Problems

To effectively apply genetic algorithms for solving robust control system design problems, we propose a novel approach called the Dimension-Reduced Sweeping Adaptive Multi-Objective Genetic Algorithm (DRSA-MOGA). This method integrates genetic algorithms with local optimization techniques to enhance overall algorithm performance and search capabilities. Through the implementation of fitness function normalization, dimension-reduced scanning based on optimal Pareto solution set searches, and adaptive fitness function adjustment mechanisms, the algorithm achieves significant improvements in both global optimization and local search performance. The algorithm employs real-coded genetic operations with tournament selection, simulated binary crossover, and polynomial mutation for population evolution. Simulation results indicate that DRSA-MOGA effectively solves problems with high approximation accuracy while maintaining solution uniformity across the Pareto front, where constraint handling is managed through penalty function methods in the fitness evaluation process.