An Accident Prediction Model For Expressway Based On Alignment And Pavement Condition

The infrastructure construction has been strengthened, with the rapid development of economy, and expressway mileage increased year by year, it is of great significance to research the safety of expressway. The article researched the correlativity of accident rate and alignment, pavement condition, and built an accident prediction for the expressway based on the alignment index and pavement condition index.The data of alignment, pavement condition, accident for Ka i Yang expressway was analysed. Because this paper research the relationship between accident rate and geometric alignment, pavement condition index, so homogeneous method was used for section division, that was ending the current section for another one based on the indexes of alignment and pavement condition, when any one of them changed, and there were 1100 sections altogether, average length of them was 227.64 m. The data of alignment, pavement condition, and accident rate was described according to section division for Kai Yang expressway, There were 1633 pieces of accident data altogether. The indexes of alignment included length of horizontal curve, radius of horizontal curve, deflection angle, radius of vertical curve, longitudinal length and longitudinal gradient. The indexes of pavement condition included pavement condition index, pavement damage ratio, riding quality index, international roughness index, rutting depth index, and skidding resistance index.The relationship between annual average daily traffic volume, planiform alignment indexes, longitudinal alignment indexes and accident rate wa s researched. The indexes with significant correlation for accident rate included AADT, radius of horizontal curve, deflection angle, longitudinal gradient and radius of vertical curve. AADT, longitudinal gradient and accident rate were positively correlated; radius of horizontal curve, radius of vertical curve, and accident rate were negatively correlated, and the accident rate was higher for concave vertical curve than the convex one. There was a quadratic parabola relationship between the accident rate and deflection angle of horizontal curve, and the safest deflection angle was between 20° and 25°.The relationship between pavement condition index, pavement damage ratio, riding quality index, international roughness index, rutting depth index, skidding resistance index and accident rate was researched. PCI、RQI、RDI、SRI and accident rate were positively correlated; There was a quadratic parabola relationship between IRI and accident rate; Accident rate was mainly concentrated in the larger DR.Traffic volume, geometric alignment indexs, pavement condition indexs and accident rate were selected as variables of the accident prediction model. A relational model between accident rate and traffic volume, radius of horizontal curve, gradient, skidding resistance index, rutting depth index was established through regression analysis. And the reliability of the model was verified by collinearity diagnostics and analysis of variance of the accident prediction model.