Research On The Dynamic Model And Rolling Horizon Algorithm For Public-use Bicycle System Vehicle Scheduling Problem

The public-use bicycle system is an important component of the urban public transportation system; It plays an important role in short-distance traffic and solves“the last mile problem”of bus transportation effectively. The overall service level of the urban public transportation could be enhanced by the public-use bicycle. At present, the phenomenon of“no-bicycle”and“no-parking”often occurs in the public-use bicycle system. So the reasonable dispatch of the public-use bicycle would be able to alleviate the imbalanced contradiction. It also can solve the problem of“no-bicycle”and“no-parking”, and improve the service level of public-use bicycle system.Based on the important project for key project of Zhejiang province“Research and Develop of the Vehicle and Road Resources’Intelligent and Dynamic Disposition System”(2009C03016-3), the public-use bicycle service sites’real-time status and scheduling method are researched.In the aspect of public-use bicycle service sites’real-time status researching, the bicycle quantity-capacity ratio is the ratio of bicycle quantity to service site capacity. The bicycle quantity-capacity ratio is the quantitative description of the public-use bicycle service sites’real-time status. It is to determine whether a service site need service or not, and the bicycle quantity-capacity ratio also determine the type and quantity of service. Then the back-propagation neural network algorithm is applied to forecast bicycle quantity-capacity ratio. The forecasting of bicycle quantity-capacity ratio is precondition to and is propitious to solve public-use bicycle system’s vehicle scheduling problem.In the public-use bicycle system’s vehicle scheduling problem aspect, the dynamic properties and fuzzy time windows of the public-use bicycle system vehicle scheduling problem are analyzed, and the bicycle sites’service demand and satisfaction are researched. Then, the model is developed in terms of the maximization of bicycle sites’satisfaction for the public-use bicycle system vehicle scheduling problem. Finally, the rolling horizon scheduling algorithm is applied to solve the dynamic model of public-use bicycle vehicle scheduling problem.Finally, based on B/S architecture, the software system of the dynamic public-use bicycle vehicle scheduling problem is design and developed by J2EE and Oracle. The experimental result indicated that scheduling method proposed in this paper has good performance in optimizing the quantity of bicycles in the public-use bicycle system. And“the returning problem”and“the lending problem”of the public-use bicycle can be alleviated.