Demand Prediction Of Shared Bikes Within The Service Range Of Bus Stations

With the rapid development of the transportation sector,carbon emissions are growing rapidly,slow traffic is gradually paid attention to,and shared bikes effectively solve the "last kilometer" problem with its convenient parking features,connecting areas inaccessible by public transport.However,the supply of shared bikes around different bus stations is insufficient or far exceeds the demand,resulting in unable to meet the travel needs of residents,or affecting the appearance of the city.In order to promote the reasonable operation management,delivery and scheduling of vehicles for enterprises and facilitate people’s travel,it is necessary to conduct in-depth analysis on the spatio-temporal characteristics and influencing factors of shared bikes within the service scope of public transport stations,have a deeper understanding of their travel rules,predict the demand for shared bikes,and provide data support for delivery and scheduling.This paper takes the data of shared bikes in Shenzhen as the research object,and mainly carries out the following work:(1)Explore the spatio-temporal characteristics of shared bikes within the service scope of bus stations.Through the analysis of the overall spatio-temporal characteristics of the study area,the bus routes needed to be studied are selected according to the overall territorial space planning,overall spatio-temporal characteristics and public transport status of Shenzhen.The service range of bus stops is determined through the policy background,the concept of the living circle,and the standard setting of station spacing.The service range conflict area is solved by creating Thison polygon.Based on the analysis of the spatial and temporal distribution characteristics of shared bikes within the service range of public transport stations from the perspectives of cycling quantity,cycling distance and inbound and outbound,it is found that the overall demand of shared bikes for receiving buses out of the station is greater than that for receiving buses in the station.The intensity of hotspots in the evening peak hours of working days is 20% lower than that in the morning peak hours.The connection demand has decreased by 40% to 50% as a whole,and the travel distance of weekday morning peak is 1600 to 1900 m,and that of evening peak is 1300 to 1800 m.(2)Analyze the influencing factors of shared bikes within the service scope of bus stations.The qualitative analysis of time factor,weather factor,POI factor and road factor,and the quantitative analysis of stepwise regression and multiple linear regression were used to determine the significant influencing factors of shared bikes in the service range of bus stations as time factor,POI factor and road factor.(3)Demand prediction of shared bikes within the service scope of bus stations.The Bi LSTM-Attention model and the LSTM-Attention model were constructed according to the historical data and significant influencing factors of shared bikes.The two models were used to predict the overall demand respectively,and it was found that the Bi LSTM-Attention model had higher prediction accuracy.MAE was 5.6% lower than LSTM-Attention,and RMSE was 10% lower than LSTM-attention.Comparative analysis of local prediction evaluation indexes found that: The LSTM-Attention model has a better effect on predicting the demand for shared bikes within the service range of hybrid,commercialpurchase-oriented and transportation facilities type bus stations,while the Bi LSTM-Attention model has a better effect on predicting the demand for shared bikes within the service range of resident-oriented and office-oriented bus stations.