For the problem that falls often seriously jeopardize to the health of the elderly,this paper designs a fall de-tection method based on EMG signals.Firstly,the feature of fuzzy entropy is extracted from the sEMG on the gastrocnemius and vastus lateralis muscle.Then,the weighted kernel Fisher linear discriminant analysis is proposed for the dataset imbal-ance problem that the number of activities of daily life (ADL)is far more than the fall,and the samples nuclear matrix is adjusted by the appropriate balance parameters.Finally,the fall is identified from walking,squat and sit down by this meth-od.The experimental results show that the method has 96. 7% fall and 99. 4% ADL average recognition rate,and is better than the other classification methods.%针对跌倒常常对老年人的健康构成严重危害的问题．本文设计了一种基于肌电信号的跌倒检测方法，首先提取腓肠肌和股外侧肌的sEMG的模糊熵特征作为特征向量，然后，针对日常活动动作类（Activities of Daily Life， ADL）的数目远多于跌倒类导致的数据集不平衡的问题，提出了加权核Fisher线性判别方法，采用相应的平衡参数来调节样本核矩阵，最终，将跌倒与行走、蹲下和坐下辨识出来．实验结果表明，该方法跌倒平均识别率96．7％，ADL平均识别率99．4％，识别结果优于其它分类方法．
展开▼
