基于多尺度最大李雅普诺夫指数的表面肌电信号模式识别

摘要

为了提高动作表面肌电信号的识别率,提出一种将最大李雅普诺夫指数和多尺度分析结合的方法.从非线性和非平稳的角度出发,引入多尺度最大李雅普诺夫指数特征,并应用到人体前臂6类动作表面肌电信号的模式识别中.首先利用希尔伯特-黄变换,对原始信号进行经验模态分解,即多尺度分解；然后利用非线性时间序列分析方法,计算多尺度最大李雅普诺夫指数；最后将多尺度最大李雅普诺夫指数作为特征向量,输入支持向量机进行识别.平均识别率达到97.5％,比利用原始信号的最大李雅普诺夫指数进行识别时提高了3.9％.结果表明,利用多尺度最大李雅普诺夫指数对动作表面肌电信号进行模式识别效果良好.%To increase the recognition accuracy of action surface electromyography (SEMG) signal, a method combining the maximal Lyapunov exponent and multi-scale analysis was proposed. Considering the nonlinear and non-stationary characteristic of SEMG, a multi-scale maximal Lyapunov exponent ( MSMLE ) feature was introduced and applied to the pattern recognition of six types forearm action SEMG signals. First step was to decomposite original signal using Hilbert-Huang transform (HHT) , known as multi-scale decomposition. Then, MSMLE was calculated by nonlinear time series analysis method. At last, eigenvector MSMLE was input into support vector machine ( SVM ) for recognition. The mean recognition accuracy reached 97.5% , which was 3. 9% greater than that obtained from maximal Lyapunov exponent of original signal. Results showed that the proposed method was effective and precise in the pattern recognition of action SEMG signals.