基于免疫神经网络的下肢假肢肌电信号识别方法的研究

A Study on EMG Signal Recognition of Artificial Lower Limb Based on Artificial Immune Neural Network

提出一种基于人工免疫神经网络的表面肌电信号模式识别方法.在对表面肌电信号进行预处理的基础上,免疫RBF神经网络模型中抗原集合作为网络的输入数据即为表面肌电信号,抗体为抗原的压缩聚类映射作为径向基函数神经网络模型的隐层中心,则网络的输出为下肢各关节的角度预测值.仿真结果表明,免疫RBF神经网能明显提高对肌电信号的识别准确率,这对于肌电假肢的控制具有良好的应用前景.

Based on artificial immune neural network,this paper proposed a pattern recognition method of surface electromyography.After pretreating surface electromyography signal,the antigen set of immune RBF neural network（surface electromyography signal） is regarded as input data and the antibodies（compression cluster mappings of antigens） are the hidden layer centers of radial basis function neural network model,and angle predictive values of lower limb joints are the network output.Simulation results show that immune REF neural network can obviously improve the recognition accuracy of EMG signals,and has good application prospect in myoelectric prosthesis control.