I. Introduction
Nowadays, people often suffer from accidental injury or sudden neurological diseases caused by weakened limb function or even the loss of local limb function [1]–[3]. The lack of upper limb function is particularly difficult for the patient's personal life. With the improvement of technology and medical science in recent years, a number of bionic rehabilitation hands as artificial limbs were introduced into medical field to help patients complete part of the action function and regain the ability to act, and the structure of the bionic rehabilitation hands are becoming more and more similar to human hands, they can basically meet the requirements of rehabilitation training. However, the development of rehabilitation hand control method is still not satisfactory. Researchers tried to control the hand by extracting and analyzing the human body's EEG, EMG, nerve and sound and other signals. Among them, sEMG(Surface Electromyography) has the advantages of strong real-time performance, good bionic property and with no trauma, which makes it become the most commonly used signal source in the control field of bionic rehabilitation hand[4].