求解一类非线性双层规划问题的改进层次型遗传算法

Improved Hierarchical Genetic Algorithm for Solving a Class of Nonlinear Bi-level Programming Problem

针对既有的层次型遗传算法不适用于求解上、下层决策空间均离散分布的双层规划问题,提出一类改进的层次型遗传算法。该算法采用同一种群分别在上、下层进化,再合并种群进行优势解挑选,形成新种群,并将之传递至另一层进行计算,直至满足循环终止条件。改进工作解决了既有的层次型算法选择压力难以调整,且只能应用于上、下层具有相同变量的双层规划问题。

Aiming at defect of existing hierarchical genetic algorithm which can not solve evenly discrete distributing bi-level programming problem of the upper and lower decision spaces, this paper presents a class of improved hierarchical genetic al- gorithm. This algorithm adopts the same group for evolution respectively on the upper and lower level, then merges popula- tion for dominant solution selection and forming a new type of population which are transferred to another level for calcula- tion until circulating terminal conditions are satisfied. The improvement work solves bi-level programming problem of exist- ing hierarchical genetic algorithm which can not adjust the selection pressure and is only applied in the upper and lower level of same variables.