Research On Traffic Accident Prediction And Application Based On Highway Geometrical Alignment And Speed

Through in-depth analysis of the survey data, three in the mountain area, Hill and plain terrain conditions, studied geometry shape, the relationship between speed and accidents, based on the negative binomial accident prediction model and sensitivity analysis of linear optimization and speed control are given recommendations, is of great practical value.From highway management department to collect a total of 23001 tra ffic accident data; on the basis of "homogeneous" on the highway accident prediction unit division, were the smallest unit 15148; from the flux observation station accident prediction unit in recent years, the years are daily traffic volume data. Integration processing over the years, the traffic accident data, geometric data, speed data and traffic data, construct a one-to-one relationship, established under different terrain conditions of highway accident analysis on Influencing Factors of database.Uses the regression analysis the method, divides the mountain ridge area, the heavy earthen mound area and the region of no relief three kind of topographical conditions, the analysis sole linear target with hundred million vehicle kilometer accident rate relations, and analyzed the combination linear target in this foundation(curved slope to combine road section, even vertical curve combines road section, coarsening slope road section and so on) with the accident rate relations, according to the above and had determined the accident rate was lowest when corresponded the linear condition, namely this article referred ideal linear condition.The application characteristics of Poisson model was analyzed, and basis for traffic accident distribution relatively discrete characteristics, Vuong statistic, alpha t statistic, AIC, BIC and other parameters were selected as model selection, the final construction of the negative binomial distribution applicable in different terrain conditions of traffic accident prediction model based on. Accident prediction results show that the accident prediction model has higher prediction accuracy for the highway with a certain length.Based on the principle of local sensitivity analysis, the sensitivity coefficients of the variables in the highway accident prediction model are analyzed. Combined with sensitivity analysis, the mean level of the growth factor of the number of accidents can be accepted. In mountainous area and hilly area highway horizontal and vertical combination section as the object of study, it is concluded that based on traffic safety of highway horizontal combined vertical sections key alignment index in the range of recommended values. In mountainous area and hilly area on the highway downhill running speed curve section peace road as the research object, is derived based on the traffic safety of Expressway on downhill curve of peace road speed control of the recommended value.