Using electromyographic signal (electromyogram EMG) to control the bionic manipulator technology, can greatly help patients with physical disabilities to provide convenience, improve the quality of life, such as some disabled people is only a broken hand, their entire arm and largely retain, their arm muscle contraction and expansion, the distribution in the muscle tissue nerve conduction of biological electrical signals, but no hand to show the transmission of these signals.These bioelectrical signals contain a lot of information that can be analyzed and used, such as the amplitude of muscle movement, the duration of contraction and so on.So, how to adopt the method of reasonable extract useful information in the original s EMG signal, the disputed by many scholars, this topic focuses on the research on the electrical control method of the manipulator, reasonable feature extraction method is used to pick up the s in the useful features of the EMG signal, for different types of gestures recognition algorithm research in depth, and explore the gestures recognition performance of the algorithm.In this way, the accuracy and real-time of motion recognition are improved, and the accurate control of mechanical artificial hand is realized.This research is helpful to promote the development of emg control methods, improve the handling performance of mechanical prosthetic hands, and is of great significance to the research of intelligent rehabilitation medical engineering and other fields.