Data Mining And Characteristics Analysis Of City Vehicles Based On Clustering Algorithm

With the exponential growth of the urban population and the number of motor vehicles,a series of problems have appeared in the urban traffic system,such as accidents,congestion and pollution,which make the urban traffic network overwhelmed.The traffic congestion caused by commuter vehicles has caused a great trouble to the residents.The traffic demand and traffic supply cannot be balanced only by increasing road infrastructure.As a result,people need more reasonable and effective means of traffic control and traffic planning to alleviate the pressure of traveling in crowded,and is the basis of traffic control and traffic planning of the road in the operation of vehicle data acquisition and the rational analysis.The analysis of travel characteristics of urban vehicles is one of the hot topics in urban traffic research,which has been widely paid attention by researchers and achieved many breakthroughs.However,it seems from the previous research results that there is still a need for further research on the use of automatic vehicle identification data for large-scale urban vehicle travel characteristics analysis.In terms of theoretical significance,this thesis uses two different clustering methods to analyze the data of commuter vehicles,which makes up for the single method in this direction at present.In terms of practical significance,the research results of this thesis can provide some basic suggestions for the management and control of road traffic system for city managers.First of all,this thesis summarizes the research on urban vehicle travel behavior from three different perspectives,sorts out the existing research progress in this direction and the problems worthy of further research,and summarizes the research purpose and significance of this thesis;Secondly,by introducing the HD bayonet system in the urban intelligent transportation system and the status quo of the bayonet system in Mianyang City,the data source used in the thesis is drawn out,and the applicability of its data analysis and preprocessing,to ensure the validity of the data set,for the following research to provide basic data support;Then,the thesis summarizes the travel chain and the origin–destination survey in the role of transportation planning,and use the processed data sets after processing to carry on t a day’s traffic travel statistics in Mianyang City,all the vehicles information records of a single day trip were sorted out,the travel chain and OD information of traffic were extracted;Finally,by extracting the characteristics of commuter vehicles in three spatial and temporal directions,the thesis uses two clustering methods,K-Means algorithm and GMM algorithm,to carry out clustering analysis on the vehicle data of the automatic vehicle identification data.The results show that both methods can effectively obtain the commuting vehicle information from the data source,but under the same characteristics,the clustering results of GMM algorithm are more stable than that of K-Means algorithm.However,the conclusions drawn by the two algorithms have been verified,and they are both in line with the commuting characteristics of urban vehicles.The conclusions can be drawn from the urban land planning,traffic management and control,traffic planning and other aspects,to provide solutions to the problems of urban traffic congestion.