驾驶员加速特性分类与辨识方法研究

Study on method of driver acceleration characteristic classification and identification

智能化是汽车未来的发展趋势,基于驾驶员行为特性的汽车控制对于实现汽车智能化具有重要意义。加速特性是驾驶员重要的行为特性,针对驾驶员加速特性分类与辨识方法进行了研究。应用驾驶模拟器设置了城市道路工况,选取驾驶员进行驾驶模拟实验,采集驾驶员操纵和车辆行驶状态数据,对数据进行滤波并提取特征值,应用k-means聚类算法进行聚类分析,将驾驶员加速特性数据分为谨慎型、一般型和激进型。基于高斯隐马尔可夫理论,根据实验数据建立加速特性辨识模型,通过实验对辨识精度进行了验证。研究结果表明,该研究方法能够对驾驶员加速特性进行合理分类,实现驾驶员加速特性的准确辨识。

Intelligentization is the development trend of the automobile in the future.Vehicle control based on the characteristics of driver behavior is of great significance for realizing the intelligentization of automobiles.Acceleration characteristics are important behavioral characteristics of the driver.The classification and identification of driver acceleration characteristics were studied.The driving simulator was used to set the urban road conditions,and the driver’s steering data and vehicle driving status information are collected.Then,k-means clustering algorithm was applied to cluster analysis,and driver acceleration behavior data was divided into caution type,general type and aggressive type.Based on the experimental data,the acceleration characteristic identification model which is based on Gaussian of mixture-hidden Markov theory was established.Validation of identification accuracy was verified by experiment.The results show that the research method can reasonably classify the driver’s acceleration characteristics,and the identification model can realize the accurate identification.