Extraction Of Vehicle Trajectory And Mining Of Key Sections Based On Traffic Bayonet Data

Since the vehicle trajectory data contains rich spatiotemporal information,it is one of the key research works in the field of intelligent transportation to explore the traffic operation state of road network through deep mining of vehicle trajectory information.Traditional trajectory mining mainly takes public transportation as the research object,and derives the operation state of the whole traffic network from the public transportation state.However,there are many differences between public transportation and non-public transportation.Ignoring the differences will affect the accuracy and robustness of trajectory mining.Meanwhile,the development and construction of intelligent transportation in small and medium-sized cities are still in the primary stage.However,most of the existing researches adopt relatively mature intelligent transportation systems in large-scale cities,ignoring the differences caused by the size of cities,which makes it difficult to directly refer and learn some methods.Therefore,based on the rich traffic bayonet data in the small and medium-sized city traffic system,this thesis studies and designs a special method of vehicle trajectory extraction and a method of polysemy path recognition in accordance with the characteristics of traffic bayonet data.By using the FP-growth algorithm,the distribution characteristics and association rules of key sections in the urban road network are analyzed in order to identify the main factors,affecting the key sections,and provide a basis for decision-making for traffic management and control measures.The main work of this paper are as follows:(1)This thesis studies data preprocessing method of traffic bayonet and analysis of traffic characteristics.Through the analysis of the original data of the traffic bayonet,this thesis design a pretreatment process and method of the traffic bayonet data.Based on this,we identify the adjacent bayonets,calculate the travel time and deal with the abnormal values,and obtain the actual network topology of the bayonets.In addition,we use mathematical statistics method to explore and probe urban traffic distribution law and characteristics.(2)This thesis studies the method of vehicle trajectory extraction for the traffic bayonet data and the method of polysemy path recognition for the polysemy trajectory segments.Determine the dynamic time threshold according to the historical traffic state changes(the traveling time of the bayonet pairs),and use the dynamic time threshold to divide the bayonet trajectory.In order to solve the problem of the polysemy trajectory segments caused by the ambiguous bayonet pair,this thesis use the TOPSIS method to identify the best one from the reasonable path set of ambiguous bayonet pairs to realize the path recognition of polysemy trajectory segments.(3)Based on association-rule mining algorithm,this article analyzes the distribution of key sections and association rules in urban road network.First,the FP growth algorithm is improved by changing the update method of header-table.Based on the improved FP growth algorithm,the key road sections are identified in the working day and non-working day,peak hours and all-day situations respectively,and the corresponding association rules are analyzed.Finally,the relevant suggestions of traffic management and control are proposed based on the mining results.