Research On Expressway Traffic Incident Detection Method Based On Fuzzy Theory And Incremental Judgment

Now, compared to the development of ordinary roads, freeways has achieved great success, and bring synchronous growth of economic and social benefits, besides, the service levels are also have a lot of corresponding extent, but with the increase in freeway mileage tends increasingly frequent,the freeway traffic incidents have also tend to more, and eventually lead to numbers of traffic issues, such as transport efficiency, traffic safety, environmental pollution and energy consumption issues,and so on. Automatic Incident Detection system come into the world,the system can determine the real-time state of traffic flow by means of the analysis of real-time traffic operation parameters, simultaneously,it can transport the information detected by the detector to the background supervisor, so that personnel can take certain measures to reduce traffic conflicts, and making the economic losses caused by the incident are minimized. Traffic incident detection system is a important component of ATMIS,it is significant for the safe operation of the entire intelligent transportation system.The detection algorithm is not only relate to the algorithm itself,but also relate closely to the parameter selection and the environment. The paper is based on the specific characteristics of the traffic incident,it also can be divided into two types of events:the effects of recurring incidents and non-recurring incidents, and specifically describe their different effects on the traffic flow. The paper analyzes the specific impact of traffic parameters to incident detection,and research the preferred method of traffic parameters.We can see the algorithms are based on test data abroad roads,with some theory analysis to the classical detection algorithm developed by scholars in other countries.They have defects in portability and adaptability.To these problems,this paper proposes an increment algorithm based on fuzzy theory. Compared with the classical algorithm, This one can determine whether it is a traffic incident, in the meantime,it can determine the specific nature of the incident, in essence,it is a kind of multi-input, multi-output detection algorithm. This algorithm is based on complex traffic flow state, using the concept of membership to describe specific traffic patterns,and the input parameters is speed and flow, and thus use the fuzzy theory to infer whether the traffic incident,besides, building the algorithm run flowchart.The paper makes the basic increment algorithm optimization,and then introduces the fuzzy logic theory, which is used to fusion,then constructs a new algorithm to improve the detection rate of the algorithm and other indicators, making the algorithm more applied and more easily to operate.At last,by means of TSIS microscopic simulation software,we can get the traffic flow data,and according to the simulation environment, pre-set global parameters and local parameters. Specific simulation include normal state and incident state. The former simulation can be drawn to the threshold which can be used to determine incident, the latter can be drawn to membership functions and fuzzy rules of fuzzy logic theory. The paper uses three indicators: detection rate, false alarm rate and the average detection time to evaluate this algorithm and the traditional algorithms to prove the superiority of the proposed algorithm.