研究并行基因算法求解双层规划问题及其在供应链物流分销系统优化设计中的应用。利用下层优化问题的KKT条件把双层规划问题转化为等价的单层规划问题,再利用并行基因算法对得到的单层规划问题进行全局优化,从而得到双层规划问题的全局最优解,最后,通过具体案例研究了上述算法在供应链物流分销系统优化设计中的应用。结果表明,并行基因算法求解双层规划,充分利用了现有计算环境的并行能力,加快了收敛速度,改善了基因算法的全局收敛性能,算法达到了实用化的规模,是一种很有应用前景的计算方法。
This paper explores algorithms for solving bi-level programming problems (BLPP) and its application to the optimal design for distribution systems of supply chain based on parallel genetic algorithms (PGAs). By utilizing the Karush-Kuhn-Tucker (KKT) condition, a BLPP is translated into a equivalent single level optimization problem. Then, the single level optimization problem obtained is globally optimized by PGAs. As a result, the global optimal solution is gained. Lastly, the application of this algorithm to the Optimal Design for Distribution System is examined by using an opti- mal location of a wholesaler, who wanted to build a distribution network. Results show that the algorithm presented takes full advantage of parallel computing, speeds up convergence and improves the performance of algorithm's global search, a practical size, so it is a promising computing method.