基于混载合乘的预约式公交路径优化问题研究

To meet the requirements of the urban passenger differences, diverse, high-quality , these kinds of customized bus and passenger reservation Dolmus - the reservation-transit came into being. Reservation-transit has considerable development prospects, but the present research theoretical research is pretty rare, and its hybrid rideshare route optimization has not been exploited. The issue firstly use data mining technology to determine the reservation request-response mechanism, then optimize the site for reservation-rideshare site by use of cluster analysis and median problem model and other methods , and finally combine different relationships rideshare site by through hybrid rideshare relationship, and by use of the idea of optimizing the design route of the vehicle routing problem to build the model and design algorithms, and then plan out reservation-transit route which not only taking the public welfare into account, but also ensuring the individual needs.The findings will have important implications for improving the practical values of the reservation-transit, but also will promot the deep study of the vehicle routing problem.

为满足城市客运差异性、多样化、高品质的需求，产生了诸如定制公交、预约合乘出租车这类新型客运模式—预约式公交。预约式公交拥有可观的发展前景，但既有理论研究甚少，其基于混载合乘的路径优化问题更是空白。本项目利用数据挖掘技术确定预约请求响应机制，采用聚类分析和中值问题模型等方法对预约式合乘站点进行选址，通过混载合乘关系组合不同站点，并利用车辆路径问题的优化思想构建模型与设计算法，进而规划出既兼顾公益性又确保个性化需求的预约式公交路线。这一研究成果对提高预约式公交的实践价值具有重要意义，同时也将推动车辆路径问题的深度研究。

混载合乘的预约式公交是一种满足客运多样化、需求高品质的新型公共交通方式。目前实际运用不多，相关研究的成果少。.本课题通过分析这种新型公交方式的适用场景与乘客的需求意向后，提出预约式公交产品，并利用网络平台采集若干预约请求。接着按照接受“大众化请求”，剔除“特殊请求”原则，将“去向趋同”的预约需求点筛选出来，采用基于时间度量的凝聚层次聚类法将“时间接近”的预约需求点聚为同一个类，形成若干个时间跨度，同时剔除人数少的时间跨度。然后对于每一个时间跨度内的预约请求点进行时空合理性分析，确定各预约请求点归属的合乘站点，再采用基于密度的DBSCAN空间聚类算法剔除距离较远、人数较少的合乘站点。最后建立了混载合乘路径优化模型，求解合乘站点与目的地之间的优化路径，设计了求解模型的禁忌搜索算法。