Improved cultural algorithm based on genetic algorithm

Knowledge about evolutionary information is not made use of effectively in Genetic Algorithm. While traditional Cultural Algorithms with dual inheritance structure converge slowly because Evolutionary Programming is chosen for the population model and only mutation operator is adopted in the population space. A novel Cultural Algorithm based on Genetic Algorithm is proposed. Four kinds of knowledge are abstracted. Simulation results on the benchmark single-peak optimization functions indicate that the performance of this method is much better than traditional Cultural Algorithms especially for the "plain functions". Aiming at multi-peaks optimization problem, Multi-Windows Cultural Algorithm and Multi-Windows Cultural Algorithm based on Genetic Algorithm are introduced. Simulation results on benchmark multi-peaks function indicates that the latter is more effective in optimization performance than the former.