驾驶员转向特性分类与辨识方法对比研究

针对汽车驾驶员转向特性分类与辨识问题,基于CarSim仿真平台对研究方法进行了初步探索。设计了转向工况仿真试验,采集试验数据,根据车辆最大横摆角速度,使用K-means聚类算法对驾驶员转向特性进行分类。在Matlab软件环境下分别采用学习向量量化(LVQ)神经网络、BP神经网络、支持向量机(SVM)建立驾驶员转向特性辨识模型,并对3种网络建立的辨识模型进行测试试验和比较。试验结果表明:3种辨识方法均具有较高的辨识精度,其中支持向量机方法在汽车驾驶员转向特性辨识方面具有一定的优势。

Study on classification and identification methods of driver steering characteristics

Aiming at the vehicle driver''s steering characteristic classification and identification, the research method is initially explored based on CarSim simulation platform. The simulation experiment of steering condition is designed and the test data is collected. According to the maximum yaw rate of the vehicle, the driver steering characteristics are classified by K-means clustering algorithm. The driver steering characteristics identification models are established by learning vector quantization (LVQ) neural network, BP neural network, and support vector machine (SVM) respectively in the environment of Matlab software. The test experiment and comparison are done for the three kinds of approaches, and the results show that all those three kinds of identification approaches have high accuracy, and the SVM method has a certain advantage on driver steering characteristics identification.