交通运输研究2024，Vol.10Issue(5)：52-63,12.DOI:10.16503/j.cnki.2095-9931.2024.05.005
基于SSA-SVM的寒区沿边公路潜在事故黑点识别
Identification of Potential Accident Black Spots on Cold-Region Border Highways Based on SSA-SVM
摘要
Abstract
A SSA-SVM based method for identifying potential accident black spots on cold-region border highways was proposed in order to enhance the safety and reliability of these highways and mitigate some accident risks in advance.Firstly,32 driving simulation test comparison scenarios were designed according to the characteristics of cold-region border highways.Vehicle operation indicator data and drivers'driving behaviour indicator data were gathered using driving simulator and eye tracker,and the differences in indicators were analyzed.Five indicators,including accelerator pedal position,brake pedal,steering wheel angle,lateral acceleration,and driver's pupil diameter,were selected to comprehensively repre-sent the potential accident risk of cold-region border highways.Then,a potential accident black spot identification model for cold-region border highways based on SSA-SVM algorithm was proposed,which utilized the efficient search ability and high accuracy of SSA algorithm to optimize the parame-ters of SVM model.Finally,the effectiveness of the proposed SSA-SVM model was verified using driving simulation test data,and it was compared with CPO-SVM and GWO-SVM models.The re-sults showed that among these three models,the SSA-SVM based model for identifying potential acci-dent black spots on cold-region border highways had the highest recognition accuracy,with a predic-tion set accuracy of 93.12%and an optimal fitness value of 0.001 41.This method is able to effective-ly identify potential accident black spots on highways in cold regions under different seasons,which can provide theoretical basis for formulating scientific accident prevention measures for cold-region border highways.关键词
事故黑点/驾驶模拟/支持向量机/麻雀搜索算法/寒区沿边公路Key words
accident black spot/driving simulation/SVM(Support Vector Machine)/SSA(Spa-rrow Search Algorithm)/cold-region border highway分类
交通工程引用本文复制引用
裴玉龙,金子微..基于SSA-SVM的寒区沿边公路潜在事故黑点识别[J].交通运输研究,2024,10(5):52-63,12.基金项目
国家重点研发计划项目(2017YFC0803901) （2017YFC0803901）
