基于隐马尔可夫模型的危险跟驰行为识别研究

Identification of Dangerous Car Following Behavior Based on Hidden Markov Model

随着汽车智能化发展,碰撞危险预警已成为主动安全技术中的重要研究内容。进行碰撞危险预警的关键是对车辆危险驾驶行为如危险换道、危险跟驰等进行高效和准确的识别与分析。基于此,本文以危险跟驰行为作为研究对象,通过对高速公路交通流特征参数的分析,提出了考虑行驶速度和车辆碰撞时间(time to collision, TTC)的危险跟驰行为分类标准及标定方法。通过使用前向–后向算法降低计算复杂度并利用Baum-Welch算法进行含隐状态的参数学习,构建了基于隐马尔可夫(HMM)模型的危险跟驰行为识别方法。结果分析和案例验证表明,本文所提出的HMM模型对于危险跟驰行为的识别精度较高,可以用于危险跟驰的碰撞预警,提升道路交通安全。

With the development of automotive intelligence, collision warning has become an important re-search content in active safety technology. The key of collision warning is to recognize and analyze dangerous driving behaviors such as dangerous lane changing and dangerous car following efficiently and accurately. Based on this, the paper takes dangerous car following behavior as the research object, and puts forward the classification standard and calibration method of the behavior considering driving speed and time to Collision (TTC) by analyzing the characteristic parameters of the highway traffic flow. By using the forward-backward algorithm to reduce the computational complexity and the Baum-Welch algorithm to learn parameters with hidden states, a method based on Hidden Markov Model (HMM) is constructed to identify the dangerous car following behavior. The result analysis and case verification show that the HMM model proposed in this paper has high recognition accuracy for dangerous car following behavior, which can be used for collision warning of dangerous car following behavior to improve road traffic safety.