Spatial And Temporal Travel Patterns And Demand Prediction Of Shared Bicycles

The first generation of shared bikes appeared in Amsterdam,the Netherlands,in1965.Today’s shared bikes are the fourth generation of shared bikes,called Dockless shared bikes.As the name suggests,it has no fixed stop,and can be borrowed and returned anywhere.Because of this convenient and fast characteristic,dockless shared bikes(hereinafter referred to as shared bikes)have developed rapidly,occupying all major cities and becoming one of the necessary tools for residents to travel.However,dockless shared bikes are a new thing after all,and they are developing too fast.Many problems have gradually emerged,such as excessive supply,imbalance between regional supply and demand,and random parking,which have seriously affected the appearance of the city and the smooth urban transportation system.In order to give full play to the complementary role of shared bikes in urban traffic,it is urgent for the government to regulate the management of shared bikes and enterprises to accurately allocate the number of shared bikes to meet residents’ travel needs.Based on the above problems,this paper studies the space-time travel rules of shared bikes and predicts their demands,which is of great significance to the planning and delivery of shared bikes.The main contents of this paper are as follows.Firstly,this paper analyzes the characteristics of shared bikes from four aspects,including the number of trips,the distance of trips,the spatial distribution and the flow direction of rides,according to more than 3.2 million cycling data provided by Beijing Mobike Bike-sharing enterprises on May 10,2017 and May 23,2017.Then,the nonnegative matrix decomposition algorithm(NMF)was used to decompose the spacetime matrix of shared bikes constructed in this paper,and five basic travel modes of shared bikes were found out in workdays,and the time series characteristics and spatial distribution characteristics of the five travel modes were analyzed.Finally USES MATLAB to build the nonnegative matrix decomposition algorithm based on BP neural network prediction model,predict the demand for Shared cycling,and respectively with the BP neural network forecasting model and short-and long-term memory(LSTM)neural network model for prediction of the results were compared,the results show the paper build a predictive model in the prediction of precision or practice is superior to other two kinds of prediction model.