提出一种提取尺侧腕屈肌表面肌电信号的短时统计特性,并将此特性预处理后作为自组织人工神经网络输入来判断运动状态,进而实现实时识别人体特定动作起始时刻的方法。用这种方法识别特定动作的运动起始时刻,省去了动作电位在肌纤维内传递以及肌纤维内部Ca^2+和ATP化学反应的时间,大大提高了人-计算机接口的时间效率。最后分析了特定动作误判出现的原因和解决办法。实验结果证明本文方法速度快、可靠性高,可在军事和竞技体育等领域得到广泛应用。
A method is given to recognize the start moment of the human-being upper limbs action. Statistical characters of the short time slices of the myoelectric signal of musculus flexor carpi ulnaris are preprocessed, then the results are given as the input of self-organization neural network to classify time slices and to find the action start time slice. Using the method to recognize the start moment of specifical movement increase the efficiency of man-computer interface, because it saves the time of action potential transfer in the muscle fiber and the time of chemical reaction between Ca^2+ and ATP in the muscle cell. Finally this paper analyzes the reasons of error estimation of certain actions. Experimental results prove that the method has high speed and reliability, thus it is used in martial and gymnastic application fields.