亲和力传播聚类在市域公路网节点重要度计算中的应用

Application of Affinity Propagation Clustering in the Calculation of Node Importance of Urban Highway Network

为了从根本上消除传统的公路网节点重要度聚类方法因各指标求和而削弱节点差异性的弊端,引入亲和力传播(AP)聚类算法对节点指标的主成分进行聚类,并转化为真实指标下的聚类结果,根据交通量预测及节点聚类结果按需布设公路线路。以某市公路网规划作为研究实例,对规划方案进行评价分析。结果表明,本文提出的公路网规划方案改善了研究区域当前路网的运行质量,并在规划年份能够与运输需求的增长相匹配。

In order to eliminate the disadvantages of weakening the diversities of highway network nodes after summation by traditional nodes importance clustering method, principal components of the indexes are clustered by affinity propagation （AP） clustering algorithm in this paper. The results are converted into final results in real indexes. Based on the nodes clustered and the forecasting traffic volume, origins-destinations are emplaced and related highway routes are put forward. A master plan of urban highway network is taken as an instance to evaluate the planning scheme. The results show that the planning scheme proposed in the paper has improved the operation situations of road network in that region, andthe planned highway network can meet the growth of transportation demand during planning period.