求解整数非线性规划结合正交杂交的离散PSO 算法

针对整数非线性规划问题,提出一种结合正交杂交的离散粒子群优化(PSO)算法.首先采用舍入取整方法,为了减少舍入误差,对PSO中的每个粒子到目前为止的最好位置进行随机修正,将基于正交实验设计的正交杂交算子引入离散PSO算法,以增强搜索性能；然后对PSO算法中的惯性权重和收缩因子采用动态调整策略,以提高算法的搜索效率；最后对一些不同维数的整数非线性规划问题进行数值仿真实验,实验结果表明了所提出算法的有效性.

Discrete PSO combined with the orthogonal crossover for solving integer nonlinear programming

For solving the integer nonlinear programming problems,a discrete particle swarm optimization(PSO) algorithm combined with the orthogonal crossover is proposed.In the PSO algorithm,each particle of the swarm is truncated to the nearest integer after the determination of its new position,and then each particle’s best position till now is repaired by using a stochastic method to reduce the rounding error.The orthogonal crossover operator based on the orthogonal experimental design is integrated into discrete PSO algorithm to make a systematic and rational exploration.The inertia weight and constriction factor are dynamically adjusted to improve the efficiency of PSO algorithm.Some numerical examples with different dimensions are carried out and the experimental results show the effectiveness of the proposed algorithm.