求解不等式约束优化的可拓遗传算法

The Extension Genetic Algorithm for Inequality Constrained Optimization

提出了求解不等式约束优化问题的可拓遗传算法.分别考虑种群中的可行解和不可行解,建立可拓关联函数对不可行解的优劣程度进行可拓评价,然后采用精英选择策略,确保每次迭代中均有一定数量和质量的不可行解被选择,从而避免种群陷入局部最优.引入了高斯变异维持种群多样性,提高算法搜索速度.通过对两个测试问题的实验和分析,验证了可拓遗传算法的可行性和有效性.

In this paper, an improved genetic algorithm based on extenics theory （EGA） for inequality constrained optimization problem is proposed. Firstly, feasible and infeasible solutions are considered separately. Particularly, extension evaluation method is used to evaluate infeasible solutions. Secondly, an elitist strategy is adopted to ensure the quantity and quality of infeasible solutions which are selected. This will be effective in avoiding convergence to a local optimum solution. At last, gauss mutation is introduced to maintain the diversity of population so that the searching speed will be improved. Two test problems are solved using the method. The results compared with those of other studies have shown the competitive advantage of our algorithm.