Performance Improvement of Hybrid Real-Coded Genetic Algorithm with Local Search and Its Applications

We have already proposed a hybrid real-coded genetic algorithm with local search (HRGA/LS) for improving the search performance of a real-coded genetic algorithm. To further improve the search performance of HRGA/LS, this paper proposes to use the blend crossover, BLX-alpha, instead of simple crossover. It is expected to find still better solutions by increasing the diversity of generated individuals. Simulation experiments elucidate the characteristics of group search of HRGA/LS with BLX-alpha, and demonstrate that the proposed method vastly improves search performance