论文部分内容阅读
为辅助上肢受损患者实现自主康复训练,促使肌肉信号与脑意识的再生通信,设计一种机械臂辅助式离散动作康复训练识别方法。受试者进行康复训练时,采集其肩关节4处肌肉群的表面肌电信号,提取时域特征,并采用BP神经网络分类算法对六种上肢肩关节动作意图进行模式识别。该方法能够准确建立表面肌电信号特征值与六种上肢康复动作之间关系映射模型,平均识别率高达90.27%。为基于表面肌电信号的外骨骼式自主康复训练系统提供一种可行的人机交互方案。
In order to assist patients with upper extremities to achieve independent rehabilitation training, to promote the regeneration of muscle signals and brain communication, to design a robotic-assisted discrete movement rehabilitation training identification method. During the rehabilitation training, the participants collected the surface EMGs of the muscle groups at the four shoulder joints and extracted the time-domain features. Then the pattern recognition of the six upper limbs was performed using the BP neural network classification algorithm. The method can accurately establish the relationship mapping model between the EMG signal characteristics and the six upper limb rehabilitation activities, with an average recognition rate of 90.27%. It provides a feasible scheme of human-computer interaction for exoskeleton self-recuperative training system based on surface EMG signals.