The Research On Three Dimensional Anti-collision Algorithm Based On Millimeter Wave Radar

In recent years,with the proposed of Internet plus Car,the development of driverless cars and smart cars are carrying on like a raging fire.As one of the key technologies of the driverless cars and intelligent vehicles,vehicle anti-collision technology is also the key technology to avoid traffic accidents,and has become the focus both at home and abroad.One of the key technologies in the vehicle anti-collision system is ranging technology.However the classical linear frequency modulation continuous wave LFMCW radar ranging method adopts FFT transform to carried out signal data processing.Calculation amount increases significantly with the increasing of sampling points,which lead to that the real-time performance of ranging system is not so ideal.To solve this problem,the cross-correlation function ranging method is introduced.Combined with the measured value of the cross-correlation function ranging method,we do FFT transform with partial effective sampling points of the signal sampling points,and more accurate distance information can be measured quickly and efficiently.The core technology of the vehicle anti-collision system is the establishment of vehicle safety distance model.Most of the existing safety distance models are based on the actual speed,but the relative velocity and relative information are not considered reasonably and the influence of road conditions,weather conditions and other complex traffic environment elements are not considered.For this,referring to the existing safety distance model,based on vehicle braking process,with the relative velocity angle,this dissertation establishes three different safety distance models according to the positive or negative of the relative velocity ?v,which considers not only the influence of relative velocity and the influence of different road materials but also the impacts of weather factors and of the driver's driving habits and reaction speed.Then,using the information detected and calculated in real-time by the vehicle anti-collision system,according to the vehicle anti-collision warning strategy,the system alarms hierarchically by sound,light and the liquid crystal display screen,to remind the driver of the car safety states,and starts vehicle brake system when necessary,which can effectively reduce collision accidents,and ensure the convenience,safety and comfort of the vehicle travel.We study the vehicle anti-collision system combined with BP neural network,which can deal with the questions that there are many factors that affect the safety distance,and the safety distance model is complicated and there are a lot of nonlinear changes.For this,the vehicle anti-collision system combined with BP neural network can meet the constant changing of the model parameters during actual traffic process and the requirement of updating the model constantly,so that the vehicle anti-collision system can update the model and the parameters in real time according to the traffic conditions.