| In order to enhance the traffic efficiency and transaction accuracy of expressway.In 2019,China identified the Electronic Toll Collection(ETC system)as the main technical solution and this technology has been deployed on a large scale nationwide network.The National expressway charging method has been adjusted from charging according to the shortest path to charging according to the actual driving path.Therefore,the accurate and efficient identification of the vehicle driving path becomes the basis for the operation of the expressway ETC system,i.e.The operation of the ETC system is highly dependent on the reliable identification of vehicle paths by ETC gantries distributed along more than 160,000 km of expressways nationwide,but the problems of detector equipment failure,communication interruption or working condition instability still lead to the lack of track for some vehicles.The traditional data verification algorithm has inefficiency and time consuming problems in verifying the huge ETC transaction data.Therefore,It is urgently required to study an efficient methods for identifying anomalies in ETC transaction data.The ETC transaction data contains the time point information at which the vehicle interacting with the ETC gantry equipmentwhile driving on the expressway.In this thesis,Clickhouse,an on-line analytical processing(OLAP)data warehouse tool,is used to build expressway data warehouse to obtain the ability to quickly process massive data.At the same time,the machine learning algorithm is used to deeply mine the information in the ETC transactions data,so as to realize the identification of repeated transaction,false transaction and missed transaction of ETC gantry.The main work of this thesis is as follows:(1)Based on the theory of dimensional modeling,the data warehouse model is established for the relevant data of expressway,such as ETC transaction data,two types of passenger vehicles and one type of dangerous vehicle,and the corresponding information of ETC gantry.At the same time,ETL operation is performed on the original data.The data warehouse has the functions of multi-source data integration and data processing,such as rapid extraction of vehicle driving path from ETC transaction data.(2)Establishing provincial expressway network topology model,and realizing the fusion of trajectory data of two types of passenger vehicles and one type of dangerous vehicle and the ETC transaction data through the data warehouse.The spatial information extraction algorithm of ETC gantry is designed to complete the spatial information of ETC gantries with missing longitude and latitude.(3)The Dijkstra algorithm based on depth constraint is used to obtain the ambiguous path of expressway.The missing detection section is quickly obtained by multi-table association method,and then the travel distance of the vehicle in the missing-detection section is estimated based on the random forest model.The actual traffic path of the vehicle is obtained by matching the travel distance with the road network distance of the section ambiguity path,so as to complete the repair of the missing gantry.(4)realizing the abnormal identification of the ETC gantry and its statistical analysis.Firstly,the repeated transaction of gantry is carried out through Clickhouse;Secondly,the gantry error detection algorithm is designed to identify the gantry false transaction;Thirdly,the missed transaction of the ETC gantry is identified by the trajectory reconstruction method.Finally this thesis identifies 27 gantries with highly miss transaction behavior and 6 gantries with highly false transaction behavior among1044 ETC gantries in the province. |
