Traffic Microparameters Perception Research Using Three-Dimensional Magnetic Sensors Deployed Along Travel Lane Markings

Traffic information collection technology is the foundation and core content to realize intelligent traffic,as well as the important prerequisite and basic guarantee for successful implementation of urban road intelligent transportation system.This article placed the three-precision magnetic detector independently delivered by our team in the lane line,acquisited geomagnetic disturbance information when vehicles pass,researched the in-depth study.of the traffic flow detection algorithm,traffic state discrimination algorithm,instantaneous speed detection algorithm,time share detection algorithm and vehicle classification algorithm according to the detected three-precision magnetic wave.Through the comparative analysis on the multiple wavelet functions,the filtering method based on fast wavelet transform-the Mallat algorithm is used.It can the effectivetly dislodge the original signal noise as well as fully retain the characteristic of original geomagnetic signal.The double-window waveform extraction algorithm is used to get the complete individual vehicle waveform,it could provide basic waveform data of vehicles for calculating traffic flow,traffic condition,vehicle speed,time share,vehicle type and other traffic information.The Vehicle Mixed Algorithm(VMA)which can distinguish between two vehicles passing the sensor at the same time is proposed based on the mathematical model of vehicle geomagnetic disturbance.It synthesizes the vehicle information of the three axes,X,Y,and Z,of the AMR sensor,which improves the accuracy rates of vehicle detection.Another algorithm developed in the paper is the Vehicle Motion State Discrimination Algorithm(VMSDA).It synthesizes the vehicle information of the Y-axis and the Z-axis and allows for the discrimination of a vehicle's operating status(driving on the left lane,the lane marking,or the right lane).The vehicle speed algorithm is proposed to get the instantaneous speed of vehicles with the information of the X-axis and the Y-axis.Another algorithm developed in the paper is the time share algorithm with the information of the Y-axis.The vehicle classification algorithm is also proposed based on the CART.It synthesizes the information of the X-axis and the Y-axis,extracts 16 time domain characteristics respectively from the X-axis and the Y-axis,forms the best decision tree based on optimal parameters and classifies vehicle types.All above algorithms are tested and verified,and also compared with the traditional traffic flow detection algorithm,instantaneous speed detection algorithm,time share detection algorithm and vehicle classification algorithm.