Research On Travel Behavior,Tolerance Characteristics And Spatio-Temporal Distribution Of Dockless Bike-Sharing

Dockless bike-sharing is gradually infiltrating into people's daily life,becoming an essential part of the trip chain.It brings both opportunities and challenges to urban traffic development,which will inevitably lead to fundamental changes in the original residents' travel and the operation and management of urban traffic.At present,the study of travel analysis and spatio-temporal distribution based on large data of sharing bike are still in the blank stage.It still needs to reveal the deep relationship between the behavior change of traveler and dockless bikesharing from two aspects of travel behavior and spatio-temporal characteristics.This paper takes dockless bike-sharing as the research object,and the development status of bike-sharing was fully analyzed.Taking advantage of the large data mining,analysis of data visualisation,it analysed the travel behavior characteristics of bike-sharing users.It also focused on the tolerance characteristics of travellers in transfer and connection stages between walking and sharing bike.Furthermore,the deep learning model is used to predict the short-term distribution of the sharing bike.The main work and innovations are as follows:(1)Through large data mining and analysis technology,the GPS data of dockless bike-sharing in Chengdu and Shanghai were excavated and analyzed.Visual analysis of usage characteristics of sharing bike has been done from the user data layer and travel data layer.The user data layer analysed the influence of brand,gender,age,user travel frequency characteristics.The travel data layer was divided by the working day and the non working day.From travel distance distribution,travel time distribution,travel time distribution,travel frequency distribution and O-D distribution,the travel behavior characteristics of bike-sharing users were analyzed and displayed by the data visualization method.(2)The emergence of sharing bike triggered a series of chain effects,in particular,the connection process(between origin-destination and sites)and the transfer process(between different sites)plays a more and more important role in public transport travel.This paper put forward the concept of tolerance perception innovatively,and analysed the tolerance perception factors in the process of transfer and connection.Data mining and analysis on the process of transfer and connection was carried out based on the walking mode and bike-sharing mode.By establishing the tolerance model based on interval fuzzy numbers,the tolerance perception changes were successfully explained during transfer and connection process,and then got the tolerance curve and threshold of spatial tolerance and temporal tolerance under different combination of elements.It can scientifically quantify the impact of shared bicycle on travel tolerance and trip structure.(3)Aimed at the characteristics of spatio-temporal variables of bike-sharing,an deep learning model based on Convolution Long and Short Time Memory Network(Conv-LSTM)was employed to predict short-term distribution.This method transformed the spatiotemporal distribution into a prediction of spatiotemporal sequence.By stacking multiple Conv-LSTM layers to form a structure of coding prediction,this paper has established an end-to-end training model.By extracting the characteristics of temporal and spatial variables such as regional bike density,regional agglomeration,regional demand and time attributes,we accurately predicted the distribution of regional bike in the next period.(4)Based on the conclusion of large data analysis of shared bicycle,from the perspective of sustainable development,the development trend of bike-sharing and the internal and external development environment were systematically analyzed.Furthermore,from the three angles of government,enterprises and users,the paper gives the countermeasures and suggestions for sustainable development.Based on this,the paper established a set of mature data theory framework for bike-sharing system based on user characteristics,travel characteristics,tolerance in connection and transfer process,and prediction of spatial and temporal distribution.