Research And Implementation Of Simulation System For Urban Traffic Flow And Traffic Accident

In recent years, with the sharp increasing of the numbers of the automobile, the traffic problems become more and more serious, such as transportation planning, traffic safety, traffic congestion, traffic control, etc. Up to now, traffic problems have become one of the most focus problems in our society. To address these problems, Intelligent Transportation System(ITS) has been proposed, and it has become one of the most hot research topics. As an important part of ITS, the simulation system for urban traffic flow and traffic accident has received much attention, and now is a important research topic in the area of ITS.Based on the background, this dissertation uses a simulation model to predict the urban traffic flow and traffic accident, and also to detect the black spots. Firstly, we analyze and summarize the overall development of researches on the simulation systems for traffic flow and traffic accident. And then we use related techniques and approaches to analyze traffic flow and traffic accident, respectively, and elaborate the advantage of data mining and neural network for being adopted in the area of ITS. Secondly, we propose several approaches to address the problems mentioned above, which are data cleaning based on the association rules, prediction based on the association rules, data clustering based on the cluster mining technology, classification based on the decision tree rules, and prediction based on the method of neural network, and then analyze and elaborate each of them detailedly. After the analysis of these key techniques, combining with the actual situation of urban roads, we design a data cleaning model based on Apriori association rules, a prediction model for traffic accident based on Apriori association rules, a black spot decision model based on DBSCAN mining approach, a predication model for traffic flow based on C4.5(decision tree algorithm) and BP neural network, respectively. In this dissertation, we also implement these models in the design of our simulation platform.Finally, we have tested the constructed urban traffic flow and traffic accident simulation platform using many testing cases. The test results strongly demonstrate that the simulation platform meets all expected system requirements.