作　　者： (王琳）; (罗旭）; (姜鑫）; (王宏）;

机构地区： 东北大学机械工程与自动化学院,沈阳110819 沈阳工程学院机械学院,沈阳110136

出　　处： 《汽车工程》 2017年第8期955-960,967,共7页

摘　　要： 本文中通过采用颈腰部生物力学和表面肌电信号相结合的方式,对驾驶员在驾驶过程中的疲劳状态进行了研究。首先,通过生物力学的计算与分析,合理地选择了能有效反映驾驶疲劳状态的生理信号采集位置,即颈6左右两侧上斜方肌和腰4左右两侧竖脊肌。然后,在利用经验模态分解算法对测得的肌电信号进行去噪的基础上,找出能表征驾驶员疲劳状态的颈腰部肌电特性参数,并对提取的特征参数(颈部复杂度、腰部复杂度和腰部近似熵)进行主成分分析,获得了两个主成分,有效保留有用信息,去除冗余信息,实现了特征参数的降维。最后,以此为自变量建立疲劳驾驶评价模型,有效提高了模型的正确率,加快了模型的运算速度。结果表明,该方法在对驾驶员正常与疲劳状态的区分上具有良好的识别效果,正确率可达90%以上。 The fatigue state of driver during driving is studied by biomechanics combined with surface EMG.signals in this paper.Firstly,through biomechanical calculation and analysis,the sampling positions of physiological signals,which can effectively reflect the fatigue state of driver,are reasonably selected.They are upper trapezius on both sides of 6th cervical vertebra and vertical rachial muscles on both sides of 4th lumbar vertebra.Then,on the basis of de-noising on measured EMG signals by using empirical modal decomposition algorithm,cervical and lumbar EMG characteristic parameters representing the fatigue state of driver are found,and a principal component analysis is conducted on characteristic parameters extracted (the complexities of cervical and lumbar muscle,and the approximate entropy of lumbar muscle) with two principal components obtained,effectively retaining useful information and eliminating redundant one,achieving the dimension descent of characteristic parameters.Finally,a fatigue driving model is built with the two principal components as arguments,and the model has high correctness and operation speed.The results show that the scheme proposed has good identification effects in distinguishing the normal and fatigue states of driver with a correct rate higher than 90%.

关 键 词： 疲劳驾驶 生物力学 肌电信号 复杂度 近似熵