Research On Interactive Control Methods Of Upper Limb Rehabilitation Robot Based On Surface EMG Signals

In recent years,with the aging population of our country continues to intensify,the number of patients with upper limb motor dysfunction caused by stroke is increasing,and people’s demand for rehabilitation medical treatment is growing relentlessly,Traditional physical rehabilitation training based on rehabilitation technicians have many problems,such as long period,large personnel consumption and limited success.Consequently,the research and application of upper limb rehabilitation robot has gradually become a hot area of rehabilitation scholars at home and abroad.In this paper,on account of the exploration of active motion intention recognition methods in rehabilitation training for patients,a human-robot interaction controller of upper limb rehabilitation robot based on Surface Electromyography(s EMG)is proposed to provide patients with a safe,comfortable and natural environment for rehabilitation training.The main research contents of this paper are as follows:(1)Based on Lagrange principle,a dynamic model of upper limb is established.On the basis of exploring the physiological structure and movement characteristics of the human upper limb,simplifying the structure of human body upper limb in a reasonable manner,includes shoulder joint,elbow joint,wrist joint in sagittal plane motion.Using Lagrange method,the interaction dynamics model of the upper limb is established,which sets the foundation for the design of human-robot interaction controller of the upper limb rehabilitation robot.(2)A General Regression Neural Network(GRNN)prediction model is proposed.By collecting s EMG,joint angles and angular velocities signals of upper limb,the mapping model between upper limb s EMG signals and joint motion is constructed,and the prediction model of Generalized Regression Neural Network is given.On the basis of motion intention recognition technology,joint angles and torques of shoulder joint,elbow joint and wrist joint of human upper limb are predicted and recognized.At the same time,the experimental results are compared with Radial Basis Function(RBF)neural network,which verify the feasibility and effectiveness of GRNN model for predicting upper limb motion signals.(3)A different training pattern of upper limb rehabilitation robot human-robot interaction control method is designed.In view of the unmodeled dynamics and other uncertain factors of the upper limb dynamics model,the adaptive human-robot interaction controller and robust human-robot interaction controller based on passivity principle are presented combining the active motion intention of patients.Depending on the ability and angles error during training of patients,the controller is divided into robot-dominant mode,the transition mode and the protection mode,which provides patients a safe and comfortable rehabilitation training environment.