中文摘要：

为解决大规模轴辐式网络规划问题，本论文提出了一个两阶段规划程序：首先，通过一个基于二维有序数据的聚类算法将全部网络节点划分为一系列恰当的区域；接下来，构建一个二次整数规划模型为每个区域进行枢纽选址，并设计了改进的贪婪取走启发式算法进行求解。与传统轴辐式网络规划决策模型相比较，这一方法将决策变量的数量从O（n2）降低为／O（n），较好地提升了网络规划模型的求解速度。我们在中国邮政速递物流公司陆路网络规划项目中的计算实践检验了该方法的有效性。

英文摘要：

To settle the problem of large scale hub- and-spoke network designing, this paper proposes a two-step designing procedure： First, divides the nodes into a series of proper regions by designing a clustering algorithm based on two- dimensional ordered data; Second, builds a quadratic integer programming model for locating these hubs, and designs an Improved Greedy Dropping Heuristic Algorithm to get the optimal solution. This approach decreases the number of decision variables from the traditional O（n2） in p-hub median model to O（n）, thus greatly speed up the solution of the large-scale network designing. Then, we apply this approach to design the express network of CPEL, and the results verify its effectiveness.