欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
针对驾驶过程中危险性驾驶行为状态的有效辨识问题,基于证据理论提出一套系统的驾驶行为险态辨识方法.在设定的显著性水平下,采用因子方差分析法,从驾驶行为状态因子中提取若干因子构建驾驶行为险态辨识特征集.在此基础上,分别采用贝叶斯模型、FCM模型、神经网络模型,构建3类驾驶行为险态辨识器,实现驾驶行为危险状态辨识.针对3类辨识器辨识结果的差异性,采用D-S证据理论,对3类模型的辨识结果予以融合,实现了驾驶行为状态危险等级的融合识别.最后结合实例予以试算,结果表明,对于危险性驾驶行为状态的误判率为1.73%,方法具有可行性.
英文摘要：
In order to effectively identify the dangerous driving behavior status in driving process,a systematic approach about driving behavior state identification is proposed based on evidence theory.After setting the significance level,a number of factors are extracted using factor analysis method,to construct dangerous driving state recognition feature set.On this basis,Bayesian models,FCM（fuzzy C-means） models and neural network model are used to construct three types of dangerous state identifier to identify dangerous driving behavior statues.Since the results of the three identifiers are different,a D-S evidence theory is adapted to fuse the three so that the integration of driving risk level is realized.Practical example test results show that the misjudgment rate of dangerous driving behavior status is 1.73% which proves the feasibility of the approach.
同期刊论文项目
同项目期刊论文
位置：