Study On The Influence Of In-Road Greening Setting On Driving Safety Based On The Experiments Of Real Vehicle People

The driving behavior of motorists is easily affected by the external traffic environment,and the key to safe driving lies in whether drivers can quickly and correctly process the information in the road traffic environment within a limited period of time.As an important component of the traffic environment,the current domestic research on the central green landscape is mostly on landscape design analysis,mostly in the qualitative analysis stage,with less quantitative data analysis.In order to study the influence of the road central green landscape on the driving behavior of drivers,this topic selects young novice drivers who are easily affected by external environmental changes as the subjects,and obtains the visual and physiological characteristics data and the test vehicle longitudinal and transverse motion data of drivers driving in the test sections with different road central green landscape settings through real vehicle driving tests,and extracts the parameters that can characterize the driving state of drivers Based on the data,we constructed a support vector machine(WOA-SVM)driving state recognition model based on the whale algorithm optimization,and designed a real-time safety assessment model for the driver’s driving state.The following research work was mainly carried out:(1)Real-world driving test.Based on the actual road conditions,we designed the realvehicle driver due to the test program,collected the driver’s eye movement data,ECG signal and EMG signal data in the real-vehicle test,and recorded the video data while driving the vehicle.The collected driver eye-movement data were processed based on the Lajda criterion.Using MATLAB and SPSS software,the ECG and EMG signals were filtered,noise reduction and their feature values were extracted.(2)Analysis of drivers’ visual and physiological characteristics under different road landscape settings.The test road sections were divided according to the presence or absence of central green landscape settings and different driving lanes.The K-means clustering method was used to improve the limitations of the traditional gaze area classification method and achieve more accurate driver gaze area positioning.On the basis of this,we analyzed four visual parameters of driver’s gaze,sweeping gaze,blink and pupil,and two physiological parameters of driver’s ECG and EMG,and specified the influence of the setting of central road green landscape on driver’s driving status.(3)Driver driving state identification and safety evaluation research.The longitudinal and transverse motion state indicators of the driving vehicle were analyzed and calculated,and correlation analysis was performed with the selected driver visual and physiological indicators to evaluate the selected parameters as effective representations of the driver driving states of different test sections.The parameters were identified by the support vector machine model,and the parameters of the support vector machine model were further tuned by the whale optimization algorithm to establish a support vector machine recognition model based on the whale optimization algorithm,and the comparison analysis proved that the model optimized by the algorithm had a higher recognition rate for the driver’s driving status on different test sections.Finally,using principal component analysis to classify the driver’s driving state in different test sections in terms of safety level,the research results show that the driver’s driving state safety is the lowest when driving in the near lane with central green landscape.Finally,by combining the WOA-SVM recognition model,a driver driving state real-time safety assessment model that can evaluate the driver driving state in real time was constructed and validated.