一种实时条件下的驾驶行为特性识别模型

The characteristic recognition model for real-time driving behavior

为了构建可用的实时交通安全预警系统,提出了一种具有2层结构的驾驶行为特性实时识别模型。为减少驾驶员所受外部因素的影响的种类,将驾驶员和车辆视作一个整体并将其考虑为一个黑箱系统。识别模型的第1层是在Newell跟驰模型的基础之上,结合车辆启停前后的迟滞现象,构建出的Newell修正模型。其输出是能够表征驾驶行为特性的3个参数。第2层模型则是以上一层模型输出为输入的较为成熟的聚类模型,并直接输出车辆的驾驶行为特性类别。为实时交通安全预警系统的构建打下基础。

In order to develop a real-time traffic safety warning system,a real-time identification model for driving behavior with two-layer structure was introduced in this study.To reduce the number of factors which can affect drivers,the driver and the vehicle were considered as a black box.The first layer of the model is based on the modified Newell model which was constructed by combining hysteresis and Newell model.Three parameters which can be used to characterize the driving behavior and identify it were extracted from this first layer.The second layer model which used the upper layer model output as input was a clustering model.The output was the result of the recognition of driving behavior.This model was a solid foundation for the construction of real-time traffic safety early warning system.