基于多种群遗传-模式搜索算法的函数优化与仿真

摘要

In order to improve the genetic algorithm’s inherent defects(poor capabilities of local searching and restrain⁃ing),a multi⁃population genetic and pattern search algorithm is proposed in this paper. The algorithm utilized strong global search ability of genetic algorithm,high⁃accuracy local optimization of pattern search algorithm,and the diversity of multi⁃popu⁃lation. Moreover,it added artificial selection operator to keep optimal valve of each population to improve the convergence capac⁃ity of the genetic algorithm,and assigned different control parameters to all sorts of group to consider local and global searching of algorithm. The test results of complex function show that the multi⁃population genetic and pattern search algorithm is better than simple or multi⁃population genetic algorithm,and it can also jump out of local optimum to conduct fast convergence. It is a kind of effective and feasible optimization algorithm.%　　为改善遗传算法局部寻优能力较差和易早熟的固有缺陷，提出一种多种群遗传⁃模式搜索算法。算法利用遗传算法的强全局搜索能力，模式搜索算法的局部寻优精度高的优势及多种群的多样性，加入人工选择算子保留各种群最优值，以提高遗传算法的收敛性，并且对各种群采用不同控制参数兼顾算法的全局搜索和局部搜索。通过对复杂函数进行仿真测试，结果表明多种群遗传⁃模式搜索算法比单独使用标准遗传算法和多种群遗传算法精度高，而且可跳出局部最优，快速收敛，是一种有效可行的优化算法。