基于主成分分析和神经网络聚类的城市坡道行驶工况研究

打开文本图片集

主题词：坡道行驶工况主成分分析SOM神经网络 聚类分析性能测试中图分类号：U469.11 文献标志码：A DOI：10.19620/j.cnki.1000-3703.2024090

Investigation ofUrban Ramp Driving CycleBasedon Principal Component Analysis and Neural Network Clustering

SongYuzhen'，Wu Zhimin’，YinXiaofeng'，LeiYulong²，LiangYiming' （1.Stitute ofAutomotive Enginering，Xihua University，Chengdu 61Oo39;2.NationalKey Laboratoryof Automobile Chassis Integration and Bionics，Jilin University， Changchun ）

【Abstract】Aiming at the issue of lacking slope information in urban driving cyclesused for vehicle performance evaluation，this paper proposes a methodfor Urban Ramp Driving Cycle （URDC）construction basedonSelf-Organizing Map （SOM）neural network.Typicalroaddrivingdata withurbanrampcharacteristicsiscollectedusing theaverage traficflow method.Afterpre-processing，thedataissegmentedintoshorttrips，and2O parametersrepresentingroadoperation characteristicsareselectedasthefeature parametersof theshorttrips.Thedimensionalityof these feature parameters is then reducedviaprincipalcomponentanalysis，followedbyclusteringtheshorttripsanalysisusinga SOMneural network. Accordingtotheprincipleofsmoothrampconnection，short trips with highcorelationareselecedtoconstructanurbanramp drivingcyclethat includesboth speedand slope information.Theresultsof automatictransmisionoperated in slope performancetestindicatethattheconstructeddrivingcyclecanreflectthedrivingcharacteristicsofvehiclesonroadwithurban ramp features，whichcanbeusedasthebenchmark driving cycleforperformance testof vehicledrivingonurbanramps.

Key words： Ramp driving cycle， Principal component analysis， SOM neural network， Cluster analysis，Performancetest

【引用格式】宋宇臻，吴智敏，阴晓峰，等.基于主成分分析和神经网络聚类的城市坡道行驶工况研究[J].汽车技术，2025（5）：47-54.SONGYZ，WUZM，YINXF，etal.Investigationof UrbanRampDriving Cycle BasedonPrincipal ComponentAnalysisandNeuralNetwork Clustering[J].Automobile Technology，2025（5）： 47-54.

1前言

目前，国内外车用性能测试基准的行驶工况多采用速度-时间曲线表达，由于缺少坡道信息，难以反映车辆在坡道特征道路上的行驶特性，进而影响车辆坡道行驶性能评估结果的准确性[1-3]。（剩余9319字）