基于K-Means聚类分析的汽车行驶工况构建  

Construction of Vehicle Driving Cycle Based on K-Means Cluster Analysis

作　　者：苏晓雯 董平军[1] 

机构地区：[1]东华大学,上海

出　　处：《建模与仿真》2022年第3期842-851,共10页Modeling and Simulation

摘　　要：为构建合理的汽车行驶工况,以给定的轻型汽车行驶数据为基础,分别运用运动学片段分析法、主成分分析法和K-均值聚类分析法对实测数据进行降维和分类,并结合相关系数法从各类运动学片段库中选取具有代表性的片段,构建反映汽车行驶特征的汽车行驶工况曲线。最后,为验证所构建的汽车行驶工况的有效性和精确性,计算作为评价体系的8个特征参数的相对误差和总体误差。结果表明,构建的汽车行驶工况曲线所反映的汽车运动特征在一定程度上可以代表数据源对应的特征,所构建的行驶工况具有有效性和精确性。In order to construct a reasonable driving cycle of the car, on the basis of a given light vehicle driving data, respectively using kinematics fragment analysis, principal component analysis (PCA) and K-Means clustering analysis of measured data for dimensionality reduction and classification, combined with the correlation coefficient method and the cumulative frequency method from the various segments in the library to select representative kinematics fragments, so as to build a curve of vehicle driving cycle which can reflect the characteristics of the car’s driving. Finally, in order to verify the validity and accuracy of the constructed vehicle driving cycle, the relative errors and total errors of the eight characteristic parameters of the evaluation system were calculated. The results show that motion characteristics of the vehicle reflected in the constructed vehicle driving cycle curve can represent the corresponding characteristics of the collected data sources to a certain extent, and this constructed driving cycle is effective and accurate.

关 键 词：K-Means聚类分析 汽车行驶工况 主成分分析 运动学片段 

分 类 号：U46[机械工程—车辆工程]