Map Matching And Residents’ Travel Hotspot Area Mining Based On GNSS Track Data

With the development of big data and positioning technology in recent years,GNSS trajectory data provides a new perspective to study the travel characteristics of urban residents.The taxi GNSS trajectory data can be used to understand the travel behavior of urban residents more comprehensively and accurately.By analyzing and determining the areas that urban residents are interested in,this paper provides a powerful reference for the scientific formulation of urban traffic planning and improvement measures.In this paper,the GNSS track data of 12,012 taxis in Chongqing City was taken as the data set.By preprocessing and track OD extraction,the correct track point data and the onoff track points of residents taking taxis were obtained.Based on the travel characteristics of residents,the travel time of working day residents in Chongqing is divided into six characteristic periods,including morning peak(7:00～10:00),afternoon peak(13:00～15:00),evening peak(20:00～23:00)and two peak periods(17:00～18:00).24:00～1:00 and night travel stage(1:00～7:00).At the same time,a variety of clustering methods are used to dig the hot spots of taxi boarding and unloading,and in-depth analysis of their travel behavior rules,so as to explore new solutions to urban traffic planning.The core content of this paper includes the following aspects:(1)Construction of GNSS taxi trajectory data processing and extraction model of boarding and unloading pointsIn this paper,the error data,duplicate data and missing fields are eliminated by Python language.In addition,the GNSS data were sequence-based with the license plate number and time field as the index,and the particularity of passenger carrying field in GNSS data was utilized to create a GNSS passenger extraction model.This model can effectively extract the information of O/D points in GNSS data,which provides a reliable data basis for the subsequent analysis and research of residents’ travel and the mining of hot spots.(2)Improved HMM-NEST map matching algorithm of hidden Markov modelIn this paper,an improved HMM-NEST map matching algorithm based on hidden Markov model is proposed to solve the problem of low accuracy of hidden Markov map matching in complex road network and low frequency sampling.By improving the Viterbi algorithm in HMM model and adding non-emitting states to Viterbi algorithm,the algorithm carries out dynamic interpolation of routes and considers the distance and direction of GNSS track points as well as the interrelation between track points.Combined with the characteristics of roads in the road network,the algorithm can ensure the accuracy of map matching algorithm after adding non-emitting states.Compared with the traditional HMM,the HMM-Nest algorithm adopted in this paper has higher accuracy and time efficiency,and is suitable for the road segments with different driving tracks,and has strong robustness.It can meet the needs of low sampling rate trajectory matching in the complex urban road network,and can provide technical support for the hotspot region extraction in the following paper.(3)Mining the spatial and temporal distribution characteristics of urban residents’ travel based on GNSS taxi trajectory dataBased on the pre-processed trajectory data of GNSS taxis,this paper extracts characteristic data such as resident travel time,resident travel volume and resident travel distance.Six travel characteristic time periods of working days in Chongqing are selected based on the time dimension.The spatial unit is the administrative division of residents’ travel,and the spatial and temporal distribution characteristics of urban residents’ travel are excavated by combining the concept of urban road network entropy,which lays a foundation for the exploration of urban residents’ travel hot spots.(4)Improved DTW-DBSCAN algorithm for hot spot miningIn this paper,the improved DTW-DBSCAN hotspot region clustering algorithm is used to extract the hotspot region of taxi pick-up and drop-off in Chongqing.The algorithm is based on the density of the residents’ pick-up and pick-up points.Combining the DTW space-time similarity measurement algorithm and DBSCAN clustering model,the density partition is integrated into the hotspot region extraction,and a density-based DTWDBSCAN hotspot region clustering algorithm is formed.Compared with K-mean and DBSCAN clustering algorithms,this method has better computing power of spatial temporal similarity,and better accuracy,stability and visualization effect in hot spot extraction.The innovation of this paper is to improve the map matching algorithm and clustering algorithm,and the HMM-NEST algorithm improves the matching accuracy and matching efficiency,so as to obtain more accurate taxi trajectory route.On this basis,the DTWDBSCAN clustering algorithm is used to analyze the taxi trajectory data,deeply understand the spatial characteristics of urban residents’ travel,extract accurate hot spots,and provide more targeted solutions for urban traffic management and planning.