On Traffic Flow Prediction And Incident Detection Of Freeway In Cities Based On Kalman Filtering

With the rapid advance of urbanization in China, the cities become larger and larger, as well as the urban population. However, the ground transportation in cities and towns appear congestion in different degree. The urban freeways are the artery of the traffic in cities, taking the main tasks of urban transportation. The traffic control and management on the freeways is the primary task of the traffic control in cities. Traffic control needs to accurately predict the traffic flows and immediately identify the accidents.According to concluding the current main methods of the traffic flow prediction and incidents detection algorithm, this paper establishes an appropriate model of traffic flow and accordingly transformation, then it can get a flexible, adaptable state and measurement equations. Besides, this paper introduces the Kalman filtering theory. Because the import and export traffic flows will affect the flows of the freeways, this paper takes the difference between the two flows as the system noise in the equation of state of the traffic flow. In the case of considering the import traffic flows, it applies the Kalman filtering optimal prediction theory to forecast the average density of traffic flow, and uses the measured data of the comprehensive automatic gridded motor vehicle license idendification system in Shenzheng (the key projects of the Guangdong provincial public security) to do the algorithm simulation, then gets the appropriate predictions results, and verifies the effectiveness and prediction.This paper proposes that it can use the Optimal Kalman filtering equations to filter the established equation of state of the traffic flow system and observation equation, doing the urban freeways incidents detection with reference to the DELOS algorithm, and verifying the feasibility and detection results of the algorithm by simulation. Based on the Kalman filtering, the freeways traffic flow prediction and incidents detection algorithm is a dynamic recursive algorithm, and the algorithm is sensitive to new values, so it is suitable for online prediction and incidents detection in traffic control of freeways. The results of this paper just provide a future parameters reference for the large, real-time traffic flow control of the freeway nowadays.