Research On Demand Prediction And Dynamic Allocation Of Shared Bikes Around Urban Rail Transit Stations

As an emerging mode of transportation,bike-sharing provides a new way of thinking to solve the "last kilometer" problem of public transportation.However,at present,the management system of bike-sharing is not yet mature,and there are many problems in the operation of bike-sharing,such as vehicle siltation and imbalance between supply and demand,which affect the development of society to a certain extent.The area around urban rail transit sites is one of the hotspots of bike-sharing,use the bike could transport interchange,commuter travel,entertainment and other different purposes,but at the same time also more prone to bike-sharing management issues,due to the spatiotemporal heterogeneity of rail transit stations,the demand and usage of bike-sharing around different rail transit stations are different,and the problems of bike-sharing around rail transit stations have been unable to be effectively solved.Taking the surrounding areas of urban rail transit stations as the research object,this paper focuses on how to accurately predict the demand for bike-sharing around the stations and achieve reasonable allocation among different stations.The content of this paper is mainly divided into the following aspects:First of all,the user’s characteristics of bike-sharing around urban rail transit sites were analyzed,the number,time,distance and influencing factors of sharing bikes around the stations were analyzed from the dimension of time and space,and the operation status of sharing bikes around urban rail transit stations and the necessity of deployment research were clarified through data mining.Secondly,based on the theory of time series prediction and neural network prediction theory,SARIMA forecasting model,LSTM prediction model and SARIMA-LSTM hybrid prediction model were set up respectively,and the model prediction process and framework were constructed to forecast bike-sharing usage around the urban rail transit sites.By selecting error evaluation indicators and comparing the predicted values of the single model and the mixed model with actual data,the superiority of the mixed model in the prediction accuracy of bike-sharing is verified,and the research results in the prediction of shared bicycles are enriched,and provide conditions for the realization of dynamic deployment of shared bicycles.Finally,in view of the temporal and spatial heterogeneity of shared bicycles around urban rail transit stations,a dynamic deployment strategy for shared bicycles around urban rail transit stations is proposed from four aspects: shared bicycle deployment methods,deployment skylights,deployment demand,and deployment process.Combining the actual shared bicycle scheduling process and based on the theory of production,sales and transportation problems,a model for the deployment of shared bicycles around urban rail transit stations with the goal of minimizing manual scheduling costs was established,and Beijing Xicheng District was selected as the dispatch unit to formulate the actual deployment plan for shared bicycles around urban rail transit stations in the area.