Research On Assisted Upper Limb Rehabilitation System Based On Flexible Mechanical Arm

At present,most of the human-machine interface of end-guided rehabilitation system is still based on pattern classification.In order to achieve the continuous and smooth control of rehabilitation aids,the estimation of joint continuous motion information has become a research hotspot in recent years.However,most of the current studies only focus on off-line training model and the training time is so long.In addition,the estimated motion information obtained by the training model is not stable and smooth.If it is directly used to control the manipulator,and manipulator assist patients in exercise,it will cause secondary injury to the residual limbs of the patients.In order to solve these problems,we not only need to unlock the EMG signal to improve the accuracy of the estimation model,but also need to re-plan the estimated information.The obtained planning curve can control the end movement of the manipulator and make the end of the manipulator move smoothly in the joint space,so as to realize the rehabilitation training of the upper limb assisted by the end of the manipulator.In this paper,for the purpose of robot-assisted upper limb rehabilitation,the continuous motion estimation information of the wrist joint is indirectly transformed into the control instructions of the end of the manipulator.A flexible manipulator end-assisted human upper limb rehabilitation training system is designed to realize the rehabilitation training of the upper limb assisted by the end of the manipulator in two-dimensional plane.The main research contents are as follows:(1)The surface electromyogram(sEMG)and joint angle signals of wrist joint during two degrees of freedom movement of upper extremity were collected and preprocessed;The surface EMG signal was extracted in various time-domain features,and the best time-domain feature extraction method was selected through experimental comparison.In order to overcome the defect that feature sequence can not accurately represent motion information and obtain more accurate and stable motion control information,a motion estimation model of feature sequence and joint angle information is constructed by support vector regression.A cuckoo algorithm based on Newton optimization with steepest descent correction is proposed to optimize the parameters of support vector regression model,which not only reduces the optimizing time,but also ensures the high accuracy of joint angle information estimation.The two DOF estimation models of upper limb wrist joint were established respectively,and realize the angle information of the two degrees of freedom can be estimated.(2)The real-time communication interface of iiwa manipulator and EMG acquisition equipment is established,which effectively realize the control of EMG data to the manipulator.And it can provide hardware conditions for EMG interface control rehabilitation auxiliary equipment.Aiming at the problem that the EMG interface directly controls the end of the manipulator,which is unstable and will produce jitter in the process of motion,this paper studies various trajectory planning interpolation algorithms of the end of the manipulator in the joint space,and proposes a relatively suitable and simple high-order piecewise function curve method which satisfies the requirements of the system.This method can transforms the estimated information into the continuous and smooth control instructions of the end of the manipulator.And it can control that The motion process of the end of the manipulator in joint space is continuous and smooth.At the same time the speed and acceleration at the starting and stopping positions of the manipulator can be zero and no jump between movement mode transformation which satisfies the condition that the end of manipulator assists human upper limb movement.(3)A flexible manipulator end-assisted human upper limb rehabilitation training system is designed.The estimated motion information reflects the motion intention of human wrist joint which as the input data of the trajectory planning module.the angle,velocity and acceleration of the planning module output curve are transformed into two-degree-of-freedom target displacement,target velocity and target acceleration of the end of the manipulator through parallel proportional electromyography control strategy,respectively,which can control the end of the manipulator in real time.It can achieve continuous motion control of the end of the manipulator,and the end effector of the manipulator can assist upper limb rehabilitation training according to the motion intention of the wrist joint.