Research On Prediciton And Scheduling Of Shared Bicycle Trips Based On Functional Area

With the increasingly serious traffic problems in China,the emergence of shared bicycles has catered to this crux,and has become an indispensable link in the urban travel chain.This kind of vehicle with greater flexibility and volatility has become a research hotspot since its inception.Especially in terms of space-time characteristics,space-time behavior,theoretical methods and technical means,it is still worthy of further study.This article analyzes in depth the various factors that affect travellers’ choice of bike sharing,and then uses the data visualization to show the use characteristics of bike sharing users based on the results of data mining,and analyzes the corresponding spacetime characteristics and behaviors from the attributes of the functional area.planning.In addition,based on the dynamic spatio-temporal distribution characteristics of shared bicycles,a new deep learning model is used to predict the OD distribution of the bicycles.Based on this,a shared bicycle air-conditioning model based on functional area division and a corresponding solution algorithm are established..The main work and innovation are as follows:(1)First of all,the corresponding data processing of the big data of bicycle sharing in Chengdu is carried out,and the characteristics of bicycle use are visually analyzed from the two levels of travel behavior and spatial function attributes.Data and feature mining will be analyzed from working days and non-working days,including the distance distribution of riding,the length distribution of distance,frequency distribution,time distribution,tidal behavior in subway functional areas and other functional areas.Through the data visualization method,the travel behavior of the shared bicycle and the potential connection with the attributes of the ribbon were also discovered.(2)Based on the potential law of the spatio-temporal characteristics and functional area attributes of shared bicycles,the concept of functional area attributes is innovatively introduced into the OD prediction model.A long-term and short-term memory network(LSTM)and a convolutional neural network are connected in parallel to construct a new spatio-temporal prediction framework for shared bicycle OD distribution.This framework realizes the simultaneous extraction of functional area correlation and time correlation.Space and time variables such as regional trip number distribution,bicycle density distribution,functional area attributes,and time stamp attributes are entered into the corresponding network structure and passed through the final The fully connected layer is weighted to achieve an end-to-end training framework that accurately predicts the arrival distribution of regional bicycles in the latter period.(3)Based on the space-time prediction results of the shared bicycle based on the functional area,based on the travel rules of the functional area and the travel characteristics under different timestamps,a model of air-conditioning for shared bicycles based on the functional area division is established,and the behavior of the bicycles in the model Constraints with scheduling behavior.Based on the analysis of the time-conditioning model,an improved ant colony algorithm is designed to solve the problem.By simulating the model and algorithm with examples,an approximate optimal scheduling scheme for each period is obtained,which proves the effectiveness of the model and algorithm.