基于驾驶事件的驾驶风格分类与识别方法研究

Research on Driving Style Classification and Recognition Methods Based on Driving Events

针对基于数据统计特征的驾驶风格分类识别方法容易忽略驾驶员在驾驶过程中驾驶风格多样性的问题,提出了一种基于驾驶事件、谱聚类与随机森林相结合的驾驶风格分类识别方法。设计试验采集驾驶数据,进行数据预处理,提取转弯事件和制动事件,经标准化和降维处理后,运用谱聚类算法分别对转弯事件和制动事件进行驾驶风格聚类。采用熵权法赋权得到每位驾驶员的驾驶风格权重,对比5种机器学习算法对驾驶风格识别的精确度,结果表明基于随机森林的驾驶风格识别精确度为92.73%,显著提高了驾驶风格识别准确率。

Aiming at the problems that,based on data statistical characteristics,the classification and recognition method of driving style was easy to ignore the diversity of driving style during driving,a classification and recognition method of driving style was proposed based on driving events,spectral clustering and random forest.Experiments were designed to collect driving data,and the data were preprocessed to extract turning events and braking events.After standardization and dimensionality reduction,the spectral clustering algorithm was used to cluster the driving style of turning events and braking events respectively.The entropy weight method was used to obtain the driving style weights of each driver,and the accuracy of five machine learning algorithms was compared for driving style recognition.Results show that the accuracy of driving style recognition is as 92.73%based on random forest,which significantly improves the accuracy of driving style recognition.