Research On Low Frequency MAP Matching Algorithm Based On Mobile Phone GPS Data

In order to obtain road network information and improve urban traffic conditions,intelligent transportation systems have become a popular research areas for scholars all over the world.The floating car technology solves the problem of vehicle information acquisition.However,in the actual driving process,the positioning information is inaccurate due to the positioning error.In order to accurately reflect the real driving state of the vehicle in the road network,the map matching technology has played an important role.At present,the development of map matching technology is divided into a map matching algorithm for vehicle implementation navigation and a map matching algorithm for background data center.The former mainly uses high-frequency floating car data(general sampling period is less than 30s);the latter is based on the background data processing and data collection costs and other considerations,and the collection of floating car data tends to be low-frequency(sampling interval is greater than 1min).The research of this paper is mainly map matching algorithm applied in the background data center based on the low-frequency data.In this paper,the existing map matching algorithm is analyzed and analyzed.Most of the map matching algorithms based on low frequency data do not take into account the matching accuracy and timeliness of the algorithm.In the process of data preprocessing,the difference between data is ignored.In the process of road segment screening,the factors such as distance,direction,speed and connectivity are not fully utilized.In the process of path matching,most algorithms consider globally,ignoring the influence of local neighbors on matching results.In this paper,a new incremental map matching algorithm is proposed based on low-frequency floating car data.The main work of the algorithm includes:(1)Preprocessing the floating car data and the electronic map data.In the process of processing redundant data,the data fusion processing method is proposed in consideration of the difference between the data;(2)Screening of candidate segments and candidate points.Taking into account the influence of speed on positioning error,the dynamic error region of the anchor point is established,and finally the candidate matching segment and the candidate matching point for each anchor point are obtained;(3)Map matching process,which takes into account the timeliness of the algorithm and the influence of adjacent positioning points on the matching result,adopts the idea of using three points as the sliding window point-by-point matching,and adopts the incremental matching method for the vehicle positioning points.,determining the matching position of each positioning point and the path between adjacent points in turn,and finally obtaining the best matching path of the entire data set;(4)In order to improve the efficiency of the algorithm,the algorithm adopts the method of segment matching to segment the data set by segmentation point,and adopts the incremental matching method for each segment to obtain the global matching path.In this paper,relevant experimental analysis is carried out from two aspects: algorithm matching accuracy and running time.The vehicle data used in the experiment was collected by the actual mobile phone app.The proposed algorithm is compared with the map matching algorithm based on spatial-temporal analysis,the map matching algorithm based on road network constraint and the map matching algorithm based on improved AOE network.The experimental results show that the incremental map matching algorithm based on low-frequency data has reached a satisfactory level in both the accuracy of matching results and the running time of the algorithm.The experimental results prove the correctness and effectiveness of the proposed algorithm.