基于单纯形法提出了一种具有全局收敛性质的算法来求解该问题.在该方法中,用下层的Kuhn-Tucker条件代替下层问题,将原二层线性规划转化为传统的单层规划问题.之后利用下层规划对偶问题可行域的顶点将该单层规划转化为一系列线性规划问题,从而用单纯形法来求解这些线性规划来得到原二层线性规划问题的解.最后,用实例验证了该方法的可行性.
Bilevel linear programming is a class of optimization with hierarchical structure. We propose a globally convergent algorithm to solving this bilevel problem. In our algorithm, replacing the lower level problem by its Kuhn Tucker condition, the bilevel linear programming is transformed into a traditional single-level programming problem, which can be transformed into a series of linear programming problem. So we can use simplex method to solve these linear programmings to obtain the globally convergent solution of the original bilevel linear programming. Finally, an example is given to illustrate the feasibilitv of the proposed algorithm.