Research On Multi-vehicle Real-time Cooperative Rebalancing Problem Of Public Bicycle System Supported By GIS

Public bicycle system as a "low carbon emissions" short-range green transport,is getting more and more popular in the world.But the main problem of "no bike to borrow" and "can't return the bike" still plagues its operation and development.Therefore,the short-term ride demand of public bicycle service point is studied while the mathematical modeling and algorithm designing is solved in order to provide a scientific basis for the intelligent rebalancing of public bicycle system,and help to break this shackles.Based on the short-term ride demand forecasting of public bicycle service points,the multi-vehicle real-time cooperative rebalancing model supported by GIS is established,and the optimization algorithm of the model are given.Finally the public bicycle intelligent rebalancing system is developed.The main work of this paper is as follows:(1)By analyzing the influencing factors of ride demand,the short-term ride demand forecasting model of public bicycle service point combined with recurrent neural network with LSTM unit in deep learning and mean OD of regional public bicycle,which forecasts the short-term ride demand of the service point,is established.The results show that LSTM has high accuracy in dealing with the demand forecasting problem of public bicycle system,which proves the forecasting method is feasible and effective;(2)In order to solve the public bicycle system rebalancing problem,the regional service index is established by setting the scheduling priority of different service points,which leads to the different situations of scheduling request.Then parameters are confirmed based on the dynamic demand characteristics of service points,and the multi-vehicle real-time cooperative rebalancing model for public bicycle system is built with the lowest scheduling cost,the most reasonable routing vehicle cooperation and the maximum gain of regional service index.Finally,the effective scheduling coefficient is put forward in order to optimize some parameters in the rebalancing model.After decomposing the dynamic rebalancing problem ofpublic bicycle system,and considering the various constraints existing in real scheduling,the genetic algorithm mixed with simulated annealing is designed.The rebalancing model is solved and compared with the actual scheduling process.The experimental results show that the hybrid algorithm can reduce the travel distance of the routing vehicle and improve scheduling efficiency.It seems that the multi-vehicle rebalancing model combined with ride demand forecasting can mostly satisfy the scheduling request of the public bicycle system;(3)Based on the public bicycle system in HangZhou,the ride demand forecasting algorithm and intelligent routing algorithm in this paper are used to design and develop the public bicycle intelligent rebalancing system,which is done in client/server mode.The pape introduced the system Development framework and main function,and finally through the Hangzhou Xiasha case application,showing the deployment of the system application process and its use.But also the system will be applied to Tonglu,Longyou and other places for the local public bicycle system to provide healthy operation to help.