基于双层隐马尔可夫模型的重型车辆行驶状态辨识方法研究  

Research on Driving Status Recognition of Heavy Duty Vehicle Based on Double-layer HMM Model

作　　者：朱天军[1] 孔现伟[2] 李彬[3] 

机构地区：[1]河北工程大学装备制造学院,河北邯郸056038 [2]中国汽车技术研究中心汽车工程研究院,天津300300 [3]康考迪亚大学机械与工业工程系高级车辆工程研究中心

出　　处：《兵工学报》2015年第10期1832-1840,共9页Acta Armamentarii

基　　金：国家自然科学基金项目(51205105);河北省自然科学基金项目(E2012402014)

摘　　要：为实时监测并发现重型车辆危险行驶姿态,在重型车辆侧翻预警系统中,采用一种基于双层隐马尔可夫模型的重型车辆行驶状态辨识方法,动态辨识重型车辆行驶状态。选取典型重型车辆行驶工况,采集相应工况数据,采用T-G检验法对数据序列进行剔除异常数据预处理。利用K-means算法设定重型车辆行驶状态的阈值。利用Baum-Welch算法对双层隐马尔可夫模型进行优化,应用优化后的隐马尔可夫模型进行重型车辆行驶状态在线辨识。辨识结果表明:该模型可准确、高效地辨识各个单一和复合工况下的车辆行驶状态,并且具有很好的实时性。In order to detect the risky driving status of heavy duty vehicle,the double-layer hidden Markov model（ HMM） is used for the rollover warning system of heavy duty vehicle,which can identify the driving status dynamically. The data is collected during the typical driving conditions,and the T-G test method is used to delete the abnormal data. The driving limit value of vehicle status is set based on Kmeans algorithm. Then finally Baum-Welch algorithm is used to optimize the proposed double-layer HMM,and the optimized HMM is used to identify the driving status in real-time. The identification results show that the proposed model can identify the driving status under every single driving condition as well as under the multiple driving conditions accurately and effectively. At the same time this model has a good real-time capacity.

关 键 词：兵器科学与技术 双层隐马尔可夫模型 重型车辆 行驶状态辨识 侧翻预警 

分 类 号：U462.2[机械工程—车辆工程]