Research On Mobile Node Prediction Of Transportation Cyper Physical System Based On Greenplum

As the key technology to support the development of next generation intelligent transportation system(ITS),Transportation cyber-physical system(TCPS)has the characteristics of deep integration of traffic information system and traffic physical system.Various data acquisition equipments and detection equipments in the traffic physical system,such as the hardware of intersection coil,speed monitoring,traffic accident monitoring,mobile terminal are constantly improving in the current traffic network,and massive traffic data is producing all the time.Massive traffic data is the valuable resource for transportation.It is the current priority to effectively store the massive traffic data.Data mining technology based on traffic big data has become the key technology to solve many traffic problems.In recent years,location-based intelligent services provide a new direction for the data mining of mobile node trajectory.Greenplum as a new generation of database engine,compared with other relational databases,has the advantage of large-scale storage,parallel processing,powerful working,low-cost hardware platform and so on,which is very suitable for big data storage.In this paper,Greenplum,a large-scale parallel data warehouse,is used as a storage and processing system for traffic big data.Based on the Greenplum data warehouse architecture,the big data platform of traffic information physical system is designed.With analyzing GPS historical trajectory data,locating query and predicting of the mobile node has important research value and practical significance.First,the hierarchical architecture of TCPS is introduced,and the characteristics of TCPS and its key technologies are analyzed in this paper.Secondly,the paper introduces the overall structure of Greenplum data warehouse,by comparing the current mainstream commercial database,analyzed the characteristics of Greenplum and the advantages in dealing with traffic big data,and design the big data platform of traffic information physical system.Third,the Python programming language of operating the Greenplum database is introduced.Based on the big data of GPS historical trajectory of mobile node and the requirement of location prediction,an improved DBSCAN algorithm is proposed,which first removing the duplicate data of the GPS history trajectory and then clustering the data of GPS history trajectory.Finally,it is realized by programming with Python,The GPS historical data denoising of mobile node is anaylzed,accuracy and efficiency of the algorithm are analyzed in detail,and the method of determining parameter of clustering radius Eps and quantity of clustering point Min Pts are intuduced.On the basis of this,withthe method of prediction sequence matching with the clustering results in the database,the system operating efficiency has been greatly improved,and the performance of the matching algorithm is analyzed by using the test data set.The simulation results show that the improved DBSCAN algorithm and the matching algorithm are superior in performance,and it can effectively predict the future position of the mobile node.