Development Of Multi-degree-of-freedom Bionic Hand Electromyography Control System

There are many people who are disabled because of amputation around world.Physical disability leads to a decline in human productivity and daily life.The multiple degree of freedom(MDOF)bionic prosthetic hand(BPH)can help the disabled to obtain functional compensation,also improving the living conditions.Surface electromyography(sEMG)is a bioelectrical signal,which is generated by the related muscle contraction during action.SEMG is a way of mind control that can help decode movements and provide correct instructions for BPH.This paper focuses on developing a MDOF biomimetic control system for BPH,and an intelligent BPH model based on sEMG using pattern recognition.In this paper,the scheme of the MDOF bionic hand muscle electric control system was proposed.The high-performance sEMG sensor and gesture recognition algorithm were designed.The self-designed software and hardware modules were built to complete the MDOF bionic hand prototype.The main work and innovations of this paper can be summarized as the following three points:(1)A distributed low-coupling MDOF bionic hand muscle electrical control system was designed.The entire bionic hand system was decoupled hierarchically,which was designed to a distributed structure of signal acquisition layer,application decision layer,and motion execution layer.The sEMG somatosensory wrist band was used the signal acquisition layer to collect the sEMG and the nine-axis attitude signal of the forearm.Then Bluetooth sends the data to the application decision layer.The app quickly decodes the action of the arm as an application decision layer and sends commands to the motion execution layer via Wi Fi.The operation on the mobile phone is very simple,and hands-free users can simply use other parts of the body to complete the simple operation of opening the mobile phone and applications.The bionic palm as the motion execution layer performs the responsion.The distributed solution using the single-chip and cell phone had reduced the cost and weight of the device,which also enhancing the real-time and accuracy of the motion recognition through the powerful capability of the mobile phone processor.(2)We had developed a portable wearing sEMG acquisition sensor,named sEMG somatosensory wristband.Considering the weak and susceptible interference characteristics of sEMG,an effective amplification filter circuit was designed.It including multi-amplifier circuits,high-pass,low-pass filter circuit and power frequency trap circuit.Furthermore,low-dropout linearregulator was used to charge for the band to ensure the sEMG somatosensory wristband can acquire low-noise and useful information.The band is small and light can be charged and discharged many times,which was designed with a Bluetooth to communicate with other devices.It can not only collect the sEMG of the eight-channel arm surface,but also the nine-axis attitude.SEMG sampling can reach 1 k Hz,and the sampling rate of the nine-axis attitude data can reach 200 Hz;(3)A gesture recognition algorithm for MDOF bionic hand muscle electrical control system was proposed.In this paper,the multivariate multi-scale entropy feature based on multivariate empirical mode decomposition was combined with the root mean square to form the eigenvector,which was input into the training network of the initial training in the long short time memory network.Then combined the nine-axis attitude signal in the classification decision layer to the initial model.The recognition result was corrected to complete the gesture recognition classifier suitable for the system.After embedding the classifier in the myoelectric control system,the recognition accuracy of the eight gestures(stretching the index finger,clenching fist,stretching the palm,OK gesture,stretching the thumb,stretching the index finger,wrist left rotation and wrist right rotation)was over 93%.