基于K-均值聚类的汽车驾驶姿态偏好研究

A Research of Vehicle Driving Posture Preference Based on K-means Clustering

驾驶姿态是影响汽车驾驶舒适性的一个重要因素,本文旨在结合驾驶员的人体测量数据定量分析其驾驶姿态偏好特征。首先利用线性尺寸参量代替人体关节角度参量来简化描述驾驶姿态,并提取相对尺寸参量以反映驾驶姿态偏好特征;在此基础上,邀请50名驾驶员进行驾驶姿态偏好测定实验,采用K-均值聚类法对3项表征上体姿态的样本数据进行聚类分析,最终划分为5类簇群,对应5种上体姿态特征,并运用三维图形对聚类结果进行可视化描述。结果表明,该方法能在排除驾驶员身材差异影响的基础上,对驾驶姿态偏好特征进行简化描述和快速区分,并结合目标用户的人体测量数据获得相关尺寸参量,为汽车座椅和转向盘布局设计提供数据支持。

Dr iving posture is one of the important factors af fecting the d r iv in g comfort of v e h ic le. This paper aims to quantitatively analyze the posture preference of drivers based on th e ir anthropometric data. F irs t ly the l in e a r dimension parameters of human body are used instead of its jo in t angle parameters to simply describe d r iv in g pos-ture. On this basis, 50 drivers are in v ite d to conduct a measuring experiment of d r iv in g posture preference. K-means clustering method is appl ied to perform clustering analysis on the sample data of three items representing the posture of upper- body. Final ly the sample data are div ide d in to f iv e clusters corresponding to f ive features of upper-body pos-ture , and the results of clustering are v isu al ly depicted by 3D p lo t. The results show that the method adopted can simply describe and fast distinguish the features of d r iv in g posture preference w h i le e l im in a t in g the effects of d r ive r ’ s stature difference, and the related dimension parameters obtained by using the human-body measured data of target users can provide data supports for the layout design of vehicle seats and steering wh ee l.