一种求解约束优化问题基于混合遗传算子的遗传算法

A Genetic Algorithm Based on Hybrid Genetic Operators for Solving Constrained Optimization Problems

摘要: 针对约束优化问题的最优解可能位于可行域边界的情况，提出了一种新的基于混合遗传算子的遗传算法。首先，在该算法中，杂交过程按可行个体和不可行个体分别进行，可行个体与最好个体杂交，不可行个体按照约束违反度的大小与可行个体杂交。其次，引入了一个基于边界变异和高斯变异的混合变异算子，其目的是促使不可行解变为可行解，可行解向边界移动。数值实验和比较结果表明了该方法的有效性。
>For the case that the optima of constrained optimization problems are often located on boundary of the feasible region, a new genetic algorithm based on hybrid genetic operators is proposed in this paper. First, in this algorithm, the crossover is executed according to feasible and infeasible individuals, respectively. A feasible point is always combined with the best-known one found so far for crossover, whereas an infeasible individual is selected according to the fitness for crossover with any feasible one. In addition, in order to make infeasible solutions become feasible ones and make feasible points move toward the boundary of feasible region, a hybrid mutation operator is presented based on boundary mutation and Gaussian mutation. Numerical experiments and comparison results show the efficiency of the method.