基于改进K-means聚类的汽车行驶工况构建方法研究  

Research on vehicle driving cycle construction method based on improved K-means clustering

作　　者：刘继隆 李乐 俞俊 LIU Jilong;LI Le;YU Jun(University of Shanghai for Science and Technology,Shanghai 200093,China)

机构地区：[1]上海理工大学,上海市200093

出　　处：《农业装备与车辆工程》2023年第5期105-109,共5页Agricultural Equipment & Vehicle Engineering

摘　　要：为提升对实际道路载荷数据典型工况提取的合理性,将采集到的车辆数据进行预处理、运动片段划分、特征参数提取并形成特征参数矩阵,使用主成分分析法对特征参数降维处理,通过孤立森林对降维后的数据进行离群点检测,并采用K-means算法进行聚类分析,根据各工况运行时间所占比例构建汽车行驶工况。结果表明,改进后的K-means聚类算法的Average Silhouette值提升了6.53%,聚类效果明显优化,所构建的汽车行驶工况的相对平均误差为5.09%,验证了该方法的有效性。In order to improve the rationality of extracting typical conditions of actual road load data,the collected speed-time series data is preprocessed,divided into motion segments,with characteristic parameters extracted and forming characteristic parameter matrix.Use principal component analysis method to reduce the dimensionality of the characteristic parameters,and isolated forest was used to detect the outliers after dimension reduction of data,K-means algorithm was used for clustering analysis,and the driving cycle was constructed according to the proportion of running time of each driving cycle.The results show that the Average Silhouette value of the optimized K-means clustering algorithm has been improved by 6.53%and the clustering effect is really optimized.Relatively,the average error of the driving conditions of the designed vehicle is 5.09%,which confirms the validity of the method.

关 键 词：片段划分 主成分分析 孤立森林 K-MEANS聚类 工况构造 

分 类 号：U467[机械工程—车辆工程] TP311[交通运输工程—载运工具运用工程]