基于脑-肌电信号的新型假肢控制模式的实验研究

Experimental study on new control mode based on combined EEG-EMG signals

目的:在传统肌电假肢控制信号中添加特定导联的脑电信号,以提升动作识别率。方法:通过预实验确立正式实验方案,在正式实验中选取1名符合实验需求的截肢患者,提取患者缺失小腿运动想象动作的同步脑-肌电信号,通过相关性分析筛选强相关性的脑电导联,并将其添加为辅助信号。在提取相关特征值之后使用基于遗传算法的BP神经网络进行分类学习。结果:在单一肌电信号中添加脑电信号为辅助控制信号后,神经网络的综合识别率从75.025%提升到了87.725%。结论:在单一肌电信号中添加脑电信号为辅助信号,可以有效提升人工神经网络对动作信号识别的准确度。

Objective To enhance the recognition rate by adding EEG signals with specific leads into the control signals of the traditional electromyogram(EMG)prostheses.Methods A formal experiment scheme was prepared with the pre experiment.In the formal experiment,one amputee was selected to extract the synchronous EMG/EEG of the missing leg movement imagination action,and then the strong correlation EEG leads were screened and the auxiliary signals were added by correlation analysis.BP neural network based on genetic algorithm was used to classified learning after extracting the related eigenvalues.Results The comprehensive recognition rate of neural network increased from 75.025%to 87.725%by adding EEG as auxiliary control signal to single EMG.Conclusion Adding EEG as auxiliary control signal to single EMG can improve the accuracy of mode of artificial neural network.