Improving And Evaluation Of OD Calculation Method Of Line Passenger Flow Based On Bus Intelligent Card Data

Giving priority to the development of public transport is an important measure to alleviate traffic congestion.Mastering the travel activities of urban residents and the travel characteristics of public transport is the basis of public transport planning and management.The basic data of public transport mainly including station passenger flow,line passenger flow and OD of public transport travel are of great significance in public transport planning and operation management.With the development of intelligent transportation system in China,the information collection and analysis system based on bus IC card system,vehicle GPS equipment and GIS geographic information system is constantly improved,which provides a new way to obtain basic bus data.In the data based on bus IC card system,only the time,route,vehicle and other information of passengers’ bus trip are recorded.Restricted by the data content,the basic bus data can not be obtained directly.How to integrate other data to calculate the passenger’s boarding and alighting station becomes the key to obtain the basic bus data.Domestic and foreign scholars have carried out a lot of research on this problem.Due to the differences between domestic and foreign urban residents’ travel activities and the development level of public transport,foreign research results can not be fully applied to China.Whether the analysis results of domestic related research are effective and reliable still needs a lot of empirical research.Based on the existing research,this paper optimizes the calculation method of boarding and alighting stations for cardholders by using the public transportation system of Panzhihua City.In this paper,the data fusion method is used to calculate the boarding station of the passengers who swipe the card.Because the GPS data of Panzhihua City does not record the status of the vehicles in and out of the station,considering the situation that the passengers may get on the train in front of the platform during the peak period,the time matching method is used to calculate the boarding station after optimizing the establishment of the vehicle arrival schedule.Then,aiming at the problem of getting off station calculation,by mining the passenger’s travel rules and improving the assumptions based on the bus travel chain,the calculation program is optimized under this method,and the calculation efficiency of the next station is improved by establishing and retrieving the neighborhood set of stations.Because this method can’t calculate all the IC card data,this paper summarizes the method based on machine learning,constructs several new features,and then uses the random forest method to calculate the remaining data.By combining the two methods,the next station of 100% swipe card records is calculated successfully.In view of the accuracy of the calculation results,this paper uses the off car card information of the segmented charging line as the real data to verify the calculation results of the fusion method based on the bus trip chain and random forest proposed in this paper at the individual level.The accuracy rate is 52.3%,and the effective rate is 76.2%;And explore the influence of the key parameters of the model(maximum walking distance threshold)on the calculation results,at the same time,analyze the calculation effect of different routes and different groups,and explain the internal reasons.At the aggregation level,the proposed fusion method is compared with the method based on the attraction right of the station.The results show that the accuracy of the two methods is good at the level of the passenger flow of getting off at the station,which can meet the practical needs of the project;In the aspect of OD passenger flow between stations,the latter has poor accuracy.The conclusion can provide a certain degree of reference value and significance for the extraction of basic bus data from intelligent bus data.