Research On Map Matching Method For Complex Road Network And Low Frequency Sampling GPS Trajectory Data

Trajectory data mining is one of the main research subjects in the field of intelligent transportation system,and map matching is the key step of trajectory data mining.The correct rate of map matching is an important factor affecting the trajectory data mining results.GPS trajectory data is the main source of trajectory data.However,under the constrains of the battery capacity of the GPS data acquisition terminal and the bandwidth of the mobile Internet network,most GPS trajectory data are acquired by low-frequency sampling,and this low-frequency GPS trajectory data will introduce higher noise and uncertainty in the map matching process.Aiming at the above problems,one map matching algorithm based on historical matching data is proposed.By introducing the matching probability model,the correct rate of map matching is effectively improved.Furthermore,using the positional relationship between adjacent trajectory points,one Hidden Markov Model(HMM)map matching algorithm based on trajectory point context information is designed.Compared with similar methods we can get a higher map matching accuracy rate in complex dense urban road network environment.The main work of this paper includes:1.The GPS trajectory data is preprocessed,including the abnormal data and duplicate data are eliminated.The road network topological relationship is constructed.The projection coordinates,the projection distance,the road travel angle,the travel angle and the trajectory angle of the GPS trajectory point are calculated.2.One map matching algorithm based on historical matching data is proposed.Using the existing historical matching data,the matching road segments are divided into several sub-road segments by the road segment nodes,and the number of historical matching trajectory points,the projection distance and the driving angle are calculated for each sub-road segment.Then the matching probability model parameters are trained by using the foregoing statistical results.The training parameters are used to preform road network matching on GPS trajectory data.The experimental results show that the proposed algorithm can obtain better map matching accuracy.3.One HMM map matching algorithm based on trajectory point context information is designed.The generation efficiency of the candidate road segment set is improved by setting the threshold of the candidate circle domain and the number of candidate road segments.The direction probability is introduced when calculating the weight of each candidate road segment in the candidate road segment set.The ratio of the distance and path distance are calculated by two adjacent track points before and after,which is as the state transition probability.Then the heuristic shortest path algorithm under the constraint of the rectangular search region is used to optimize the calculation of the state transition probability.The experimental results show that,the proposal algorithm can obtain better correct accuracy than the similar HMM algorithm for the map matching of low-frequency GPS trajectory data in the complex dense urban road network environment.