Research Of Road Safety Based On Human And Environment Factors

The development of the transportation industry and mass construction of thehighway road have made great contribution to the progress of the society and theliving of human race, however，the corresponding accidents also brought lots ofcatastrophes. It is of vital importance to arrange lots of concerns on the research ofroad safety. Based on systematic analysis of the reasons of accidents, this paper doessome researches on the development trend of the fatigue, sleepiness, reaction time anddriving speed with the extension of continues driving time under real drivingcircumstance, and investigates the effects of road and environmental variables toaccidents frequency relying on Ontario vehicle accidents data.The main work and innovative conclusion is illustrated as follow:1. This paper analyze the relationship of explosive factors, such as fatigue,sleepiness, reaction time and driving speed, with continuous driving time under realdriving circumstance. The statistics analyses show that it is strongly recommended tohave a rest for the occupational driver after continuously driving3hours.2. In light of previous research that focusing on human factor and ignoring theroadway geometries, weather, and other non-behavior factors, a multi-variables ofhighway accidents is analyzed and a negative binomial model is applied to quantifythe effects of non-behavior factors to frequency of occurrence of highway accidents.Elasticity analysis is used to measure the sensitivity of frequency of occurrence ofhighway accidents to non-behavior factors. By studying the relationship betweenweather and geometric elements, this paper quantified the effects of the non-behaviorfactors to road safety.3. According to the constraints of the traditional log-linear function forresearch of the variables and accidents frequency, that the elasticity is constant, adynamic elasticity function form widely used in the economics context is applied tocompare with the traditional log-linear function form. It is found that, in terms ofgoodness-of-fit, the translog function performs better than the traditional NB model.The dynamic elasticity of translog model is more approach the real circumstance.