提出一种改进差分进化算法求解混合整数非线性规划问题。该算法利用同态映射方法,解决差分进化算法无法直接处理整数决策变量问题;提出改进的自适应交替变异算子,提高算法的搜索性能;提出一种自适应保留不可行解的方法处理约束条件,并对差分进化算法的选择算子进行改进,提出一种直接处理约束条件的新选择算子。六个常用的混合整数非线性规划问题的实验结果表明了该方法的有效性和适用性。
This paper proposed an improved differential evolution algorithm for the mixed integer nonlinear programming problem.Used a homomorphous mapping in the algorithm to deal the integer decision variables,and used a self-adaptive alternating mutation operator to improve the search performance.In addition,it used a self-adaptively infeasible solution biased way to construct a new selection operator which could deal the constraints directly.The experiment results of six benchmarks show the new algorithm is efficient and practical for the mixed-integer nonlinear programming problem.