基于主成分和聚类分析的汽车行驶工况构建

Construction of Vehicle Driving Cycle Based on Principal Component Analysis and Cluster Analysis

基于城市汽车实际道路行驶采集的数据,运用主成分和K-means聚类分析,对汽车行驶工况进行构建并分析误差.首先,对5种类型的原始数据,依次进行时间不连续数据、尖点数据、毛刺噪声数据以及怠速异常数据等预处理;其次基于预处理后的数据,对运动学片段进行划分,同时引入14个相关的运动特征指标,采用主成分分析进行降维处理并选取了具有代表性的5个主成分,基于K-means聚类分析算法将其运动学片段按照拥堵路况、一般拥堵路况和通畅路况分为3类,再根据聚类结果选取最优工况片段,从而得到最终的工况曲线.最后,对汽车行驶工况与所采集数据源的各运动特征值进行误差分析,验证了所构建的汽车行驶工况的合理性.

Based on the data collected by the actual roads of urban vehicles,this paper builds and analyzes the driving conditions of vehicles based on principal component and K-means cluster analysis.Firstly,four types of the original data are sequentially subjected to four preprocessing such as time discontinuity data,cusp data,glitch noise data and idle abnormal data.Secondly,based on the preprocessed data,the kinematic segments are divided.At the same time,14 related motion feature indicators are introduced,and principal component analysis is used to reduce the dimensionality.Five representative principal components are selected.Based on the K-means clustering analysis algorithm,the kinematic segments are classified according to the traffic conditions and general road conditions.And the smooth road conditions are divided into three categories,and then the optimal working condition segments are selected according to the clustering results,thereby obtaining the final working condition curve.Finally,the error analysis of each indicator value of the driving condition of the vehicle and the data source collected by the city is carried out,and the rationality of the driving condition of the constructed vehicle is verified.