求解非线性双层规划问题的混合变邻域粒子群算法

针对非线性双层规划难以获得全局最优的问题,汲取粒子群算法的快速搜索能力及变邻域搜索算法的全局搜索优势,提出了求解非线性双层规划问题的混合变邻域粒子群算法.首先利用Kuhn-Tucker条件,将非线性双层规划转化为一个单层规划问题,然后由粒子群算法得到一个较优的群体,通过审敛因子判断陷入局部最优的粒子,并进一步利用变邻域搜索算法的全局搜索能力对陷入局部最优的粒子进行优化,从而得到全局最优.测试函数的仿真实验对比分析证明了该算法的有效性.

A hybrid intelligent algorithm by combining particle swarm optimization with variable neighborhood search for solving nonlinear bilevel programming problems

In this paper, a hybrid intelligent algorithm by combining the particle swarm optimization (PSO) with variable neighborhood search (VNS) is presented on the basis of analyzing the problem of nonlinear bilevel programming. This method integrates the fast search capability of PSO with the global search ability of VNS. Firstly, the bilevel programming is transformed into a single level programming problem by use of the Kuhn-Tucker conditions. Then, the preferable swarm is obtained by PSO algorithm. Furthermore, the swarm get into local optima, which is estimated by convergence criterions, is optimized by VNS algorithm. Finally, the result of benchmark problems demonstrates the proposed algorithm is effective than the compared algorithms.