A Study Of Public Bicycle Repositioning Problem

With the development of the city, the level of motorization becomes higher and higher. As a consequence, problems such as traffic congestion, environmental pollution become more and more serious. In such a situation, the concept of green transportation with low-energy, low-emission, low-pollution is accepted by more and more people. Public bicycle has an advantage in alleviating traffic jams and solving the ’last mile’ problem. But due to the unreasonable development plan of public bicycle system and inefficient repositioning, travelers always be rejected by the lack of bicycle or locker at a depot. This problem restricted the harmonious development of public bicycle system. So our study is to overcome this problem.Firstly, this paper provides a comprehensive literature review about this problem both at home and abroad. After that, this paper introduces the characteristics of public bicycle and emphasizes the necessity of repositioning. At the same time, we analyze each part of public bicycle system such as the pattern and cost. As for the vehicle routing problem, we introduce the weakness of traditional pattern which divides the stations into several sets and one vehicle serves one set. On this base, we put forward a conception of multiple stations cooperation.Secondly, this paper proposes an optimization model for the public bicycle repositioning problem so as to minimize the total repositioning cost. In the model, we take all station into consideration at the same time, and analysis the optimal path for each vehicle. We propose a hybrid heuristic algorithm to solve the model, and the hybrid heuristic algorithm contains Genetic Algorithm and Tabu Search Algorithm. First, using genetic algorithm to obtain the public bicycle station sets which will be repositioned by different vehicles. Then using hybrid heuristic algorithm to get the optimal path of each vehicle.Finally, we take the public bicycle system in Zhongshan as a case to conduct the application of our repositioning model. Through the comparison of two results calculated by our model and traditional model, our model is proved to be more effective.