Highway Traffic Unusual Event Based On Data Fusion Detection

The detection of traffic incident has all been difficulty and significant in the monitoring and controlling system of freeway all the time. Existing traffic incident detection system is unsatisfactory in practical application. Data fusion technology is a new kind of information comprehensive processing method in recent years, already there are a lot of tries in ITS at present. Based on analyzing the existing automatic incident detection of freeway traffic, combining data fusion theory principles with system architecture, this paper analyses and studies freeway unusual incident detection based on the.data fusion by MATLAB software.This paper constructs the traffic incident detection system, and introduces the four subsystems in detail : information collection subsystem, information transmission subsystem, information processing subsystem, information issue subsystem. On the basis of the system, this paper proposes method of multisensor data fusion, which fuses loop coil vehicle sensor, video vehicle sensor, weather sensor.In data low-grade processing stage, this paper utilizes wavelet analysis theory for processing the data collecting by loop coil vehicle sensor and compares with three methods effects.In data high-grade processing stage, this paper illuminates the two kinds of data fusion systematic model of traffic incident detection: systematic model of centralized fusion method, systematic model of anechoic distributed fusion method. Using the simulation data trains LVQ network and ameliorated BP network, then detection effect is validated by actual data.