采用小波变换进行肌电信号预处理与多尺度分解,并采用小波系数最大值与平均能量值作为肌电信号特征,采用支持向量机进行特征分类识别的运动解码,并用此方法进行了腕部动作识别的实验.与时域特征、频域特征、AR参数特征提取方法以及神经网络识分类别方法进行对比,结果表明:基于支持向量机的小波特征提取方法可以较好地区分不同腕部动作,具有最高的分类精度,极大改善前臂假肢的操纵性能. The wavelet transform was used to pre-process and decompose EMG signals. The maximum and average value of wavelet coefficients were used as the EMG signal characteristics, and the SVM was used to perform feature classification and motion decoding. The wrist motion Compared with time domain feature, frequency domain feature, AR parameter feature extraction method and neural network classification method, the results show that the wavelet feature extraction method based on SVM can better distinguish between different wrist movements, With the highest classification accuracy, greatly improve the performance of forearm prostheses.