基于随机森林算法的轿车碰撞损伤预测研究

打开文本图片集

中图分类号：TN911.7-34；TP181 文献标识码：A 文章编号：1004-373X（2025）16-0139-07

Research onvehicle collision damage prediction based on random forest algorithm

LIU Xin， LIU Conghao，LI Gang，AN Xunan，TONG Shiyu， SUN Yilong （SchoolofAutomotiveandTransportationEngineering，LiaoningUnversityofTechnology，Jinzhouooo，China）

Abstract：Inorder tominimize thedegreeofdamage suferedbyvehiclesinunavoidablecollsionscenariosand improve roadtrafficsafety，avehiclecollsiondamageprediction modelbasedonthecollsionsimulationdatasetandtherandomforest algorithmisproposed.Thefiniteelementmodelofvehiclecolionsimulationisestablished，andthevehicledamagedataset under16Osetsofwokingcondiionsisobaindbyangingtheollier，collisonangle，collisonoffet，collisonsdetc. This datasetisused toestablishanavehiclecolision damage prediction model basedontherandom forest algorithm，nd the damagepredictionforautomobilecollisionsisprformed.Themultipletestingresultsshowthatthecollsiondamageprediction modelhas an averageabsolute percentage errorof20.09%androot meansquareerorof 3394.Incomparison withthe support vectormachinepredictionmodel，therandomforestcollsiondamagepredictionmodelhasbeterfitingeffect，andthedegreeof dispersion between the predictedandreal values islower，whichcan more acuratelypredict the specificdamagevaluesof the keypointsofthevehicleafteracollsion.Itcanprovideamoredetailedandaccuratereferenceforintelligentdrivingvehicle trajectory planning systems and adaptive constraints，so as to improve road safety.

Keywords：vehiclecollisiondamageprediction；randomforestalgorithm；collisionsimulationdataset;collsioncondition setting;trajectoryplanning;adaptive constraint

0 引言

目前，车辆智能化技术快速发展，已经在汽车领域取得显著成果。（剩余7034字）