Research On Perception Method Of Driving Safety Distance

In the past 20 years,the rapid growth and popularity of automobiles have made China the largest country in the global automobile production and sales.The number of traffic accidents and casualties has also increased year by year,which makes our country the traffic safety problem is getting more and more serious,and it also puts more stringent requirements on the traffic safety technology of traditional vehicles.Driving safety distance sensing is an important part of the vehicle's active anti-collision technology.Its research is very important for improving the vehicle's active safety technology and reducing the accident rate.This paper first analyzes the background of driving safety distance perception prediction,the research status at home and abroad,and the research significance of the topic.In view of the advantages of support vector machine(SVM)in solving nonlinear,high-dimensional and local minimum values,a support vector regression(SVR)method is proposed to predict the driving safety distance.Then based on the SVR research,the square of the training error is used instead of the slack variable,and the inequality constraint is replaced by the equality constraint.Then the least square support vector regression(LS-SVR)driving safety distance prediction model is proposed.In this way,the secondary planning problem of the solution can be avoided,and the training speed of the model is greatly improved.However,LS-SVR loses the looseness and robustness of SVR,which affects the accuracy of the model to some extent.In order to avoid the above problems,a method of driving safety distance prediction based on weighted least squares support vector regression(WLS-SVR)is proposed.In addition,since the reasonable setting of model parameters is critical to the accuracy of prediction results,the particle swarm optimization(PSO)algorithm is used to optimize the parameters in WLS-SVR,and a prediction method based on PSO-WLS-SVR is proposed.The prediction accuracy of the model is further improved.Finally,this paper uses the driving test sample data obtained by driving safety distance simulation software to compare and analyze the simulation results of three prediction methods based on SVR,WLS-SVR and PSO-WLS-SVR.The experimental results show that the prediction result of WLS-SVR method is better than SVR method,and the prediction error of PSO-WLS-SVR prediction method is the lowest among the three methods,which indicates that PSO parameter optimization method has certain improvement in model prediction accuracy.In addition,it also shows that the PSO-WLS-SVR method has certain effectiveness and superiority in driving safety distance perception prediction.It also shows that the weighted least squares support vector regression prediction method based on particle swarm optimization has certain research value and society significance.