Analysis And Prediction Of Commercial Vehicle Drivers’ Reaction To Active Safety Warning System

With the rapid development of society and economy,vehicles have become an indispensable tool in our life.The rapid development of vehicle numbers also brings serious traffic safety problems.In all traffic accidents,the percentage and severity of commercial vehicles are much higher than private cars.Therefore,more and more active safety warning systems have been applied in commercial vehicles to improve drivers’ decision-making ability.This article took commercial vehicles as targets,firstly,analyzed drivers’ subjective attitude on active safety warning systems by using questionnaires.After that,large amounts of naturalistic driving data were applied to further analyze the characteristics of forward collision warning、headway monitor warning and lane departure warning data.A random forests algorithm was adapted to evaluate the relationship between driver acceptance and relevant factors.In addition,a driver reaction prediction model was proposed using GA-BPNN to predict the reaction of commercial vehicle drivers using various key factors.The detailed contributions of this article are summarized as follows:(1)Drivers’ subjective attitude on active safety warning systemsThe objective attitude of commercial vehicle drivers was summarized by questionnaires.Some statistical methods were applied first to describe the basic information of drivers and some working scenarios of active safety warning systems.A Likert scale was used to further analyze the subjective attitude of drivers.Totally four common factors were summarized using factor analysis.At last,the attitudes of commercial vehicle drivers were quantified via fuzzy synthetic evaluation method.In general,the commercial drivers’ attitudes were positive.But the positive degree is not so high as expected.(2)Active safety warning data characteristic analysis51 days’ naturalistic driving data of 24 commercial vehicles were used in this article.The characteristics of these data were analyzed.Analysis of variance was conducted to evaluate the significance of driver reaction and relevant factors under different types of warning functions.The results showed that drivers have more rapid reactions to FCW signals,the reaction to HMW and LDW signals are slighter.The results of ANOVA indicated that the key factors under FCW and HMW signals are duration time,driver age,road type,vehicle type,warning time,and vehicle speed.In addition,the key factors under LDW signals are duration time,road type,weather condition,driver age,and warning time.(3)Analysis of key influence factors on driver reactionThree Random forests algorithms were applied using the data of three types of warning functions,separately.The key influence factors under different types of warning functions were obtained.The OOB estimate of three models were 0.816,0.771 and 0.820,the fitting degree is relatively high.The result showed that three key influence factors under forward warning collision warning system are vehicle speed,duration time and warning time;Three key influence factors under headway monitoring warning system are vehicle speed,driver age and warning time.;In addition,three key influence factors are duration time,vehicle speed and driver age under lane departure warning signals.The differentiated design of three types of active safety warning systems were discussed base on these key influence factors.(4)Prediction of driver reaction under the influence of warning signalsConsider the inner relationship between driver reaction and various influence factors under FCW,HMW,and LDW signals,the article applied GA-BPNN to establish three driver reaction prediction model under different active safety warning signals considering different driver characteristics and different road and environment conditions.By examining the goodness of fit of these three models,all of them have a good performance and don’t have over-fitting problems.In addition,the prediction speed and accuracy of GA-BPNN were all better than BPNN.The model in this article can be applied in the prediction of commercial vehicle drivers and give advices to the improvement of active safety warning system design.