OD Estimation Of Bus Passengers And Optimization Of Bus Network Based On Multi-source Data

For the development of a city’s transportation,the development of urban public transportation services is a very important part of it.Due to its unique advantages,urban public transport has long been a necessary mode of transportation for urban residents.In recent years,the automatic toll collection system for public transportation and the automatic vehicle positioning system have been widely used.A large amount of public transportation IC card data and public transportation GPS data have been produced,which contains a large amount of passenger travel information that is waiting to be mined.Compared with the traditional manual survey method,the use of multi-source data fusion method to collect passenger travel information can save a lot of manpower and material resources,and can extract more comprehensive travel information to provide a scientific basis for urban bus operation and management.Based on this,through the integration of bus IC card data,bus GPS data and corresponding bus line and station information,this paper proposes a method for estimating bus passenger boarding stations based on secondary clustering using the DBSCAN algorithm.By setting space and time thresholds to judge the transfer behavior of traveling passengers,a three-mode fusion calculation method is formulated for the calculation of alighting stations,and a bus network optimization model is designed and solved by an ant colony algorithm.The specific content includes:First,the three types of data sources for multi-source data fusion mining are explained,namely,the data sources of bus IC card data,bus GPS positioning data,and bus station static location data.At the same time,the data formats of the three types of data sets are analyzed and studied.The method of data preprocessing.Secondly,in order to improve the accuracy of the boarding station estimation,this paper proposes a method for obtaining boarding stations based on the idea of secondary clustering.This method uses the DBSCAN algorithm to perform two-density clustering.The first is to locate the GPS trajectory data.The purpose of clustering is to remove the noise points in the GPS data;the second is to cluster the IC card check-in location with the bus stop location to obtain the specific station where passengers board the bus and check-in.Based on the data set of passenger boarding stations,this paper studies the calculation method of getting off stations.The calculation of the drop-off station can be understood as two stages.First,the transfer behavior in the passenger travel data must be recognized.Secondly,three modes of drop-off station calculation methods are developed for non-transfer data,respectively,for closed trips.A drop-off station identification method based on travel chain rules of the chain,and a drop-off station identification method based on personal or historical travel behavior for single or broken links.Finally,this article is carried out according to the calculation method of getting on and off stations.The theory of optimization of the public transportation network is also explained,and the setting method of the objective function and restriction conditions of the optimization of the public transportation network is explained.And through the obtained bus passenger OD data,combined with ant colony algorithm to solve and analyze the optimization model.