Application Of Data Mining Technology In The Intelligent Traffic

In recent years, with the accelerating process of urbanization and the rapid development of social economy, the number of private cars has increased significantly, and the traffic pressure has increased rapidly. Frequent traffic accidents threaten people’s lives and the serious traffic congestion brings people inconvenience. The traditional solutions, such as widen roads and traffic control, have great limitations which can not meet the current traffic demands. At present, our country is vigorously developing the smart city construction, and smart transportation is an important part of this plan. Data in the traffic field has the characteristics of large quantity, high dimension and complex type. Achieving effectively traffic flow forecast according to data mining technology has important research value and practical significance for easing traffic jams and Information benefiting.The scientific prediction of short time traffic flow is an important part in the implementation of smart transportation system. The significant characteristics is high degree of uncertainty and nonlinear because traffic system is participated by many people at the same time and is various from time, location and other factors. This causes a big difficulty in traffic flow prediction. The more interval prediction time is, the more probability of unpredictable sudden accident events. Thus, comparing with the long-term prediction of traffic flows, short-term traffic flow forecasting has more practical meanings and becomes the hotspot of research in recent years.This thesis firstly analyzes characteristics of urban traffic flows, and emphasizes the necessary of accurate descriptions. It also states relevant technologies and specifically explores the actual demands and limitations of public traffic flow prediction based on actual road network data. At the same time, it makes key analysis on different road sections and provides preliminary validation for actual feasibility of applying traffic flow prediction theory. According to the concepts of urban short-term traffic flow BP artificial neural network prediction method and public traffic flows, with analyzing actual data, it is found that public traffic speed in previous adjacent section at T has a good correlation with traffic flow in this section at T+10 minutes. Moreover, large and medium cities have basically formed the main transit network pattern and modern Intelligent bus system system could timely collect accurate data of traffic flow operation information, which provide the basis of urban short-term traffic flow BP artificial neural network prediction method based on operation states of public traffic.This thesis puts forward an urban short-term traffic flow BP neural network forecasting method based on public traffic flow operation state and analyzes the relevant network model, parameter configuration, prediction process and so on. With actual data in Zhangyang Road and Zaozhuang Road in Shanghai Pudong, the research builds an urban short-term traffic flow BP neural network forecasting model based on public traffic flow operation state, and makes tested and contrast on conventional urban short-term traffic flow forecasting model. The result shows the prediction value in new method is more closely to the actual value. Thus, method proposed in this thesis is more practical and valuable.