Evaluation on low-grade highways roadside safety in mountainous area by projection pursuit model

In the recent years, the roadside safety accidents happen frequently in our national mountainous area, therefore, determining the roadside safety grades scientifically and reasonably is imperative for the prevention of more roadside accidents in the mountainous area. To qualitatively evaluate the roadside safety grade of the highways in the mountainous area, this paper adopts the Friedman-Tukey projection index to establish the projection pursuit model and introduces the multi-agent genetic algorithm to solve it. After that, the weights of factors influencing the roadside safety of the highways in the mountainous area can be acquired. Then, the Fisher optimal dissection method is applied to group the samples according to their projection values. Therefore, the number of grades and corresponding threshold values can be obtained. Finally, take highways in the mountainous area of Luanping County as an example, the reasonability and the validity of the established model were verified.