The Research Of Finger Rehabilitation Training System Based On Multichannel Continuous Myoelectric Control

Nowadays,it’s common that the function of the hand is damaged because of some diseases such as stroke and trauma.During the traditional rehabilitation therapy,the patients’rehabilitation training completed with manually assistants from staffs.Thus this method costs too many medical resources.It will be a very promising way if patients can finish the finger rehabilitation training exercises by themselves with the exoskeleton rehabilitation system,which can not only increase patients’ participation experience but also improves the patient’s self-rehabilitation training effect.Myoelectric signal(EMG signal)has become the regular source for controlling of the rehabilitation system due to special characteristics of non-invasive,well real-time performance and usage of reporting the body movement information.However How to obtain the surface myoelectric signal and the pattern recognition of myoelectric signal are the difficulties in rehabilitation system research.The current EMG-controlled rehabilitation system widely uses off-the-shelf,costly EMG acquisition instrument to collect surface EMG signals.Then process the collected signal and control the rehabilitation equipment by computer.However,this method is only suitable for the study but not for the commoditization of EMG control Rehabilitation system.This paper analyzes the research status and application of finger rehabilitation training system at home and abroad.It proposes a finger rehabilitation training system based on multi-channel continuous myoelectric control.The specific research work is as follows:(1)According to the characteristics of surface EMG signal,EMG signal conditioner,EMG signal processing/controller were independently developed to build EMG signal acquisition system,which can complete a series of functions,such as amplification,filtering,A/D conversion,data transmission,processing and rehabilitation control.(2)According to the generation mechanism of myoelectric signal,the non-invasive metal electrode was selected for the collection of the surface EMG signals,and the electrode placement arm cover was designed to fix the metal electrode.According to the muscle distribution of the forearm and the type of the rehabilitation action,the position of the electrode was determined to ensure the stability of the signal acquisition.(3)According to BP neural network theory and the characteristics of time domain,the various parameters of BP neural network model were determined,the classifier based on the BP neural network was designed.The square demodulation method was proposed to extract the characteristic envelope of the original surface EMG signals and complete the formulation of the teacher’s sample label.(4)The artificial skeleton rehabilitation hand was designed in which the joint actuator uses the high reliability spiral cylinder.According to the characteristics of pneumatic system and myoelectric control,a multi-point continuous electromyography control method was proposed.A system experiment platform was established to complete the training experiment of the finger rehabilitation system by myoelectric control.The experiment shows that the design of the finger rehabilitation training system based on multi-channel continuous myoelectric control has a higher real-time and correctness,which is helpful for the patients’ rehabilitation training.