Research On Scheduling Problem Of Shared Bicycle Considering Dynamic Demand

With the development of Mobile Internet and Mobile Payment and the rise of Sharing Economy,Shared Bicycle emerges as the times require.It effectively deals with the "last mile" problem,fills a gap in the market,solves the pain point of public transportation,and meets the needs of users because of its high convenience.Therefore,Shared Bicycle has grown rapidly these years.However,the shared bicycles are often uneven distributed during rush hour because Shared Bicycle has no parking pile.With the tightening of the policy,it is impossible for Shared Bicycle companies to satisfy the demand of bicycles in the areas whose demand is high by simply increasing the number of bicycles in the future.The uneven distribution of bicycles can only be balanced through reasonable scheduling.This paper is based on the actual order data of Mobike.The scheduling of Shared Bicycle is cited as the study object.The division of bicycles scheduling areas and the prediction of bicycles demand are explored.The Shared Bicycle scheduling model with dynamic demand is established.On this basis,the scheduling scheme of Shared Bicycle in rush time is discussed,so as to achieve the balance between scheduling cost and service level.Firstly,according to the order data of Mobike,this paper analyzes the characteristics of bicycles using in different kinds of regions and the influencing factors of bicycles using.It is found that the bicycles in the university area have obvious feature of uneven distribution.Therefore,the surrounding area of a certain University is selected as the research area.There are three peak times of bicycles using in one day:morning,noon and evening.Morning 7 am to 9 am is selected as the research period in this paper.Based on the Shared Bicycle order data around the University,DBSCAN is used to divide the research area into several scheduling areas and K-means is used to get the location of demand points in each scheduling area,which provides location information for the later algorithm.Then,this paper constructs prediction model of ride in quantity and ride out quantity based on the order data by using the LSTM.The prediction results of the model accord well with the actual data.The model is used to get the prediction of ride in and ride out quantity of each area in the morning rush hour,which provides the demand data for the later algorithm.Finally,this paper considers the time window,vehicles capacities and other constraints to establish the Shared Bicycle scheduling model under the condition of dynamic demand.This paper also designs a genetic algorithm to solve the model.The algorithm is tested by Gurobi and the result is good.Based on the location of demand points and the demand of each area,the changes of scheduling time window and the penalty cost of demand loss are made to get the shared bicycle scheduling schemes under different conditions and these scheduling schemes are compared with each other.It provides a reference for the selection of appropriate scheduling scheme.There are 29 figures,13 tables(17 including appendix),68 references in this paper.