一种快速遗传算法及其收敛性

针对遗传算法局部搜索能力较差、全局优化速度缓慢、搜索的效率对约束惩罚因子的选择有明显依赖性等问题,设计了能够从可行解空间和不可行解空间同时搜索的交叉算子、具有前期的快速搜索能力和后期的最优解保持能力的变异算子和具有"精英"保持能力的选择算子,并采用已搜索解集避免了子代的"返祖"和退化现象。在此基础上,设计了具有全局搜索能力的遗传算法,并证明了算法的收敛性。仿真结果表明,该算法能够快速地找到全局极值点。

Fast genetic algorithm and its convergence

Aiming at problems of genetic algorithm,such as a worse local search,a slower global optimization,and search efficiency depending on the selection of penalty function obviously,a crossover operator which searchs from both feasible and infeasible solution space simultaneously,a mutation operator which can rapidly search prophase and hold global optimal solution anaphase,and a selection operator which can hold optimal solution are designed.Moreover,the reversion and degradation are avoided via using the searched solution space.Based on all these,a novel effective genetic algorithm for global optimization is proposed and its global convergence is proved.At last,the simulation result shows that this algorithm can rapidly find the global extremum point.