Research On Travel Trajectory Extraction Method Based On Mobile Phone Data

The travel trajectory data of individuals in the city can provide a large amount of information for research and work on urban job-living balance,traffic optimization,information census,etc.,thereby promoting the convenience of urban residents’ commuting and reducing motor vehicle travel.With the rapid development of communication technology,mobile phone data is gradually applied to the study of travel trajectory extraction.This article studies the travel trajectory extraction method based on mobile phone to further apply mobile phone to job-residential balance and urban Traffic research provides higher-quality data.The main research contents are as follows:First,by analyzing the positioning principle of the mobile phone data,the error source and error types of the mobile phone data are explored.The original mobile phone data id preprocessed,and different types of error data processing methods and parameters are given.Secondly,by analyzing the temporal and spatial distribution characteristics of mobile phone data,a travel trajectory extraction model for which based on segmented travel chains is proposed: a hierarchical clustering algorithm based on dynamic temporal and spatial thresholds is proposed to analyze the stay points of the preprocessed data,Extract the OD points in the travel chain,thereby dividing the travel chain into multiple segments of single travel.Based on a precise integrated transportation network,a single travel mode is preliminarily determined.In a single travel,some algorithms based on graph theory are applied to generate a travel trajectory candidate set,and the travel trajectory is further determined by the spatio-temporal trajectory similarity algorithm.Thirdly,the hierarchical clustering algorithm based on dynamic spatiotemporal threshold and the travel trajectory extraction model of mobile phone data based on segmented travel chain are used as the experimental group,and the clustering algorithm based on density and hidden Markov path extraction algorithm are used as the control group.In addition,the experimental results are compared with the actual trajectory to verify the effectiveness of the model.The experimental results show that: compared with the clustering algorithm based on density,the proposed hierarchical clustering algorithm based on dynamic spatiotemporal threshold can more accurately identify "pseudo trip" and more accurately divide multi segment trips;Compared with hidden Markov model,the mobile phone data travel trajectory model based on segmented travel chain can extract user travel trajectory more accurately.In the case of driving travel or public transport travel with complete data,the accuracy of trajectory extraction is more than 90%.