基于图论求解多选择背包问题.pdfVIP

( ) 计算机工程与设计 Computer Engineering and Design 3144 2009,30 13 计算机工程与设计 Computer Engineering and Design 人工智能 基于图论求解多选择背包问题 1 2 李炯城 ， 鲍江宏 (1. 广东省电信规划设计院有限公司信息系统研究院，广东广州510630 ； 2. 华南理工大学数学科学学院，广东广州51064 1) 摘 要：多选择背包问题涉及的约束条件种类最多，在背包问题的各种变形中最为复杂。使用动态规划的思想，巧妙地把这 个组合优化领域的问题转化成图论上求最短路径的问题。因为标准的Dijkstra 算法只能找出两个节点间的一条最短路径，为了 克服这个问题，对该算法进行了改进。对案例的测试表明，该算法能成功地算出多选择背包问题的全部最优解。首次把动态规 划、图论算法共同应用到多选择背包问题，既能发挥动态规划的理论优势来大大减少计算量，又能充分利用图论的已有成果。 关键词：多选择背包问题; 图论; 最短路径算法; 动态规划; 多阶段决策过程图 中图法分类号：TP301.6 文献标识码：A 文章编号：1000-7024 (2009) 13-3144-04 Solution to multiple-choice knapsack problem based on graph theory 1 2 LI Jiong-cheng , BAO Jiang-hong (1. Information System Research Institute, Guangdong Planning and Designing Institute of Telecommunications Limited Company, Guangzhou 510630, China; 2. School of Mathematical Sciences, South China University of Technology, Guangzhou 51064 1, China) Abstract ：Among variations of Knapsack problem, multiple-choice Knapsack problem (MCKP) is the most complicated, for it is involved in the most types of constraint condition. First, the problem from combinatorial optimization is skillfully transferred to the shortest path problem from graph theory on the basis of dynamic programming. Because the standard Dijkstra algorithm can only find one shortest path between two vertexes, the algorithm is correspondingly improved. Final, it is tested by the case, and turned out to be able to find all optimal so