Classifiable Automatic Incident Detection Methods Based On Data Fusion

At present, the urban traffic congestion is not only seriously affected the daily work of urban residents and urban life, but also hindered the economic, cultural, political, and many other fields. Can solve this problem is related to the urban development potential and international competitiveness. The reason why the traffic jams on the road is because the road there is the "traffic bottleneck" which reduce the traffic flow in the space constraints, the occasional traffic congestion is the main reason for causing "traffic bottlenecks",. At the same time, the main reason caused by road traffic incident is the occasional traffic congestion, therefore, how to timely detect and effective deal with the traffic incidents is very concerned by traffic management , and it is also the focus of this paper.Now, with the rapid development of detector technology, traffic incident detection data may come from different types of detectors, such as induction loop detector, ultrasonic detector, video detector, and the GPS floating car detector, which requires the complementary among the data of different detectors to be considered during the traffic incident detection process. Data fusion technology has a good capacity that can combine the advantages of different data, which provide the necessary technical means to improve the effectiveness of traffic incident detection for use multiple data sources. In addition, various types of traffic incident has the different effects on the traffic flow, the impact of similar incidents that happens in traffic peaks or traffic low peak is also different, which requires a traffic incident grading. Therefore, there are more adequate objective conditions to do the in-depth study for classification traffic incident detection methods based on the data fusion.According to relevant research situation, combined with the author's point, this article expands the analysis and research about automatic traffic incident detection based on data fusion methods from the following aspects:1) It lays the information for the study of classification traffic incident detection methods based on data fusion to analyze the characteristics of traffic flow and traffic information collection method.2) Traffic function model of data fusion is presented based explaining the definition and levels of data fusion. It selects the data fusion method for this article by analyzing the characteristics of the main flow data fusion method and needs of this study.3) for objective reality that all types of traffic incident have different impact on the traffic flow and similar traffic incidents at different times have different effects on the traffic flow, from the perspective of enhancing ease traffic flow, the new method of classification of traffic events based on the traffic impact is proposed.4) A classification in the traffic event for traffic incident detection algorithm based on the design of artificial neural network method is proposed based on of re-classification. It verifies its effectiveness and comparative analysis using the measured data of a large city.By the above, the study aimed to provide more decision-making information for traffic managers to facilitate the timely processing of traffic incidents that is even more serious impact on traffic flow, so as to solve the traffic congestion caused by traffic incidents.