基于PCA主成分分析和K-means算法的汽车行驶工况数据量化研究  

Quantitative Research on Vehicle Driving Condition Data Based on PCA Principal Component Analysis and K-means Algorithm

作　　者：王沛[1] 陈劲杰[1] WANG Pei;CHEN Jinjie(University of Shanghai for Science and Technology,Shanghai 200093,China)

机构地区：[1]上海理工大学

出　　处：《软件工程》2020年第3期17-21,共5页Software Engineering

摘　　要：随着我国经济的快速发展,从汽车大国到汽车强国的逐步转变,汽车数量也急剧增加。本文针对轻型汽车实际道路行驶采集的数据(采样频率1Hz),处理为各个运动学片段,采用PCA结合K-means++聚类方法,对处理后数据样本进行降维处理,分析其中主要特征成分,将各运动学片段依据综合特征指标归类,计算主要特征参数,使用相关系数筛选典型特征片段。构建典型汽车行驶工况曲线。使用K-means聚类处理数据段,计算处理结果并分析与总体样本特征偏差范围,判断工况曲线构建的合理性,是否符合世界WLTC工况标准。结合汽车标准行驶工况比较分析综合特征指标差异。With the rapid development of the Chinese economy,the number of cars has also increased dramatically,since the gradual transformation from a large automobile country to the car power.This paper focuses on the data collected from the actual road driving of the light vehicle(sampling frequency 1Hz),then processes the data into each kinematic segment.Using PCA combined with K-means++clustering method,the processed data samples are subjected to dimensionality reduction processing.Then the main characteristic components are analyzed.Each kinematic segment is classified according to the comprehensive feature index.Then the main feature parameters are calculated.Lastly,the correlation feature is used to filter the typical feature segments.The typical vehicle driving condition curve is constructed.The K-means cluster is used to process the data segments.The processing results are calculated,and the deviation range from the overall sample characteristics is analyzed to determine the rationality of the construction of the working condition curve and whether it meets the world WLTC working condition standard.The characteristics and difference of the comprehensive characteristic indicators are compared and analyzed in combination with the standard driving conditions of the automobile.

关 键 词：PCA K-means++聚类 汽车标准行驶工况 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]