一种求解随机线性二层规划问题的分支定界-粒子群混合算法

将一类随机线性二层规划模型转换为带期望约束的确定性线性二层协方差规划模型,并进一步利用KKT条件将二层协方差规划模型转化为单层规划模型,然后利用分支定界-粒子群混合算法对该模型进行求解.与传统分支定界算法的对比实验表明,该算法有效改善了上层问题的方差结果,且计算效率得到了较显著提高.

A Particle Swarm Optimization-Branch and Bound Algorithm for the Linear Stochastic Bilevel Programming Problem

First, a class of stochastic linear bilevel programming models is transformed into a deterministic linear bilevel covariance programming model with expected constraints. Then, using the KKT condition, the deterministic bilevel covariance programming problem is transformed into a deterministic single level programming problem. After that, the single level programming problem is solved by the particle swarm optimization-branch and bound algorithm. The results of a comparative experiment show that the proposed algorithm can improve the variance results of the upper level problem and significantly increase the computational efficiency.