Spatio-temporal Analysis And Similarity Measure Of Floating-car GPS Trajectories

Taxi GPS trajectory data contains a lot of information,and fixed data format is easy to analyze.In addition,the taxi route and travel time are completely determined by the passengers.Studying the movement and behavior patterns of taxi passengers is helpful to make live more convenient.Trajectory similarity measure is the basis and key problem of trajectory data mining.Therefore,the method and application of trajectory similarity measure based on Beijing taxi GPS trajectory data are studied in this paper.1.Research on GPS trajectory similarity measure method.A trajectory similarity measure method based on direction characteristics as the leading factor and taking into account the distance feature is proposed.The distance feature is measured by surrounding area,and the direction feature is measured by real average direction,linear average direction and longest common direction sequence.Firstly,the combination of distance feature and direction feature is used to evaluate the performance and verify the reliability.Then combination of the surrounding area and the linear average direction is selected as the trajectory similarity measure.Next,the validity and accuracy of the combination between surrounding area and linear average direction are verified by comparing with classical trajectory similarity measure LCSS(longest common subsequence).The results show that the method has a high degree of discrimination in the linear fitting trajectory direction by clustering GPS passenger trajectory data of Beijing taxi.It can be indicated that similarity measure can better characterize the trajectory trend direction.2.Application research of mobile trajectory data based on POI data.Firstly,the OD(origin destination)point of the passenger segment trajectory is extracted,and the POI attribute is matched to each point by using the Baidu map API.Then select the pick-up and drop-off point for the four periods from May 11(working day)and May 16(weekend)for 8-10 hours and 20-22 hours.And Spatio-temporal analysis was conducted on trajectory classification at starting areas,such as residential areas,shopping,food,and hotels.It was found that the trajectory of working days and weekends was similar in terms of driving distance characteristics and time characteristics.According to the statistical results of interest types,the main types of starting point(residential areas,shopping,food,hotels)and the destination(residential area,hotel)at20-22 o'clock on weekdays and weekends are the same,and the changes in taxi passenger activity are the transition to residential activities with high similarity.The types statistical results of the two periods in the same day show a certain difference,the type of getting on and off in the morning of the working day is very extensive while the nighttime period is relatively concentrated in the type of residence.The main starting points types of the weekend are the same,but the end point is different.The morning session has more types than the night session,and the morning session includes the tourist,airport entrance and exit attractions that are not available on the working day.The distribution of interest points in each period was compared and analyzed,it was found that the two periods in the same day showed a wide distribution range and large area in the morning session,and the distribution was concentrated at night.