基于数据特征的驾驶风格分类与识别方法研究  

Driving Style Classification and Recognition Method Based on Data Features

作　　者：刘冠颖 郭凤香[1] 申江卫[1] 沈世全[1] 陈峥 LIU Guanying;GUO Fengxiang;SHEN Jiangwei;SHEN Shiquan;CHEN Zheng(Faculty of Transportation Engineering,Kunming University of Science and Technology,Kunming 650500,China;Department of Public Basic Disciplines,Yunnan Open University,Kunming 650500,China)

机构地区：[1]昆明理工大学交通工程学院,云南昆明650500 [2]云南开放大学公共基础教学部,云南昆明650500

出　　处：《昆明理工大学学报（自然科学版）》2023年第3期165-173,共9页Journal of Kunming University of Science and Technology(Natural Science)

基　　金：云南省自然科学基金项目(202101AU070034,202101BE070001-058);云南省教育厅科学研究基金项目(2023J0796)。

摘　　要：对汽车驾驶员驾驶风格的有效识别能够保障汽车安全行驶,提升车辆燃油经济性.提出了一种基于改进K均值聚类和多类支持向量机的驾驶风格分类识别方法,并对驾驶员在驾驶过程中不同时间表现出的驾驶风格进行分类识别.首先,对采集到的驾驶数据进行了有效清洗、离散化处理及数据特征提取,将数据特征集划分为训练集和测试集;然后对训练集中的怠速数据单独给定标签,其他数据运用主成分分析和改进K均值算法进行驾驶风格聚类,在此基础上,采用多类支持向量机、逻辑回归、K近邻等学习方法对驾驶风格进行识别;最后,将识别模型用于测试集数据以测试其泛化性能,研究结果表明,该模型的识别准确度超过96%,表明构建的驾驶风格分类与识别模型具有较高精度.The effective identification of driving style can guarantee the safety of vehicle,and improve the fuel economy of vehicle.In this study,a method based on improved K-means clustering algorithm and multi-class support vector machine is proposed to identify the driving style of drivers in the driving process.Firstly,the collected driving data are effectively cleaned,discretized and the data features are extracted,and the data features set is divided into training set and test set.Then,the idling data in the training set are given separate labels,and the other data are clustered using principal component analysis and the improved K-means algorithm.On this basis,multi-class support vector machine,logistic regression,k-nearest neighbor and other learning methods are used to identify the driving style.Finally,the recognition model is used to test the generalization performance of the model by testing the number of sets.The results show that the identification accuracy is up to 96%,which indicates the high accuracy of the driving style classification and recognition model constructed in this paper.

关 键 词：驾驶风格 数据特征 K均值聚类 多分类支持向量机 主成分分析 

分 类 号：U463.6[机械工程—车辆工程] TP311.13[交通运输工程—载运工具运用工程]