Research On Key Technologies Of Upper Limb Rehabilitation Exoskeleton Robot System

With the development of society and the aging of the population,the number of patients with upper limb motor dysfunction which is caused by stroke is increasing,and the corresponding demand for rehabilitation training has also increased significantly.Rehabilitation training for stroke motor function injury mainly revolves around a range of motion and muscle strength training.Traditional rehabilitation training has many problems,such as a huge workload for doctors,a high dependence on doctors’ experience,and a large number of repetitive training actions.As a new type of rehabilitation equipment that can assist patients with upper limb motor dysfunction in rehabilitation,the upper limb rehabilitation exoskeleton can assist doctors to provide patients with continuous effective and multi-modal rehabilitation training,At the same time,it can also strengthen the active patient’s awareness of active training,and provide an objective basis for the evaluation of the patient’s rehabilitation effect and the improvement of the treatment plan,which is of great significance for alleviating the shortage of rehabilitation medical resources.However,existing rehabilitation exoskeleton robot devices generally have problems such as poor portability and comfort caused by inadequate structural optimization design,poor human-machine coordination caused by inaccurate motion intention recognition,and poor joint motion control effectiveness caused by insufficient smoothness and stability of joint motion trajectory tracking control.In this paper,a 7-DOF upper limb rehabilitation exoskeleton robot is designed to address the shortcomings of the existing rehabilitation exoskeleton robot.The main contents of the paper are as follows:Based on the clinical rehabilitation theory of upper limb motor dysfunction and the anatomical and kinematic characteristics of the human upper limb,and referring to the needs of rehabilitation training safety,individual difference adaptability,and training mode diversity in the use of upper limb exoskeleton robots,the upper limb rehabilitation exoskeleton was designed,which mainly include the design of shoulder joint,elbow joint,wrist joint motion module,and hardware control system.By designing three rotating joints connected in series with an acute angle to each other to form the degree of freedom of the shoulder joint of the upper limb,the range of motion of each joint is increased while avoiding the occurrence of singular point positions.Through the design of planetary gear combined with parallelogram mechanism to realize the inner/outer rotation degree of freedom of upper limb elbow joint.A kinematics model of an exoskeleton robot for upper limb rehabilitation based on the screw method is proposed and its workspace is analyzed.Aiming at the problem of determining the extreme working conditions of the parts to be optimized in existing topology optimization methods for the lightweight design of upper limb rehabilitation exoskeletons,a new topology optimization method based on the orthogonal experiment method to determine the extreme working conditions of the parts to be optimized(Orthogonal Experiment-based Working Conditions Topology Optimization,OEWC-TO)was proposed.The orthogonal experiment is designed to determine the joint angle combination of the robot when the parts to be optimized generate the maximum stress,so as to determine the corresponding extreme working conditions of the robot,on which the topology optimization of the parts is based.Then,the lightweight design of the parts is realized with the minimum mass as the optimization goal and the end deformation as the constraint condition.The lightweight comparative simulation of the OEWC-TO method and the topology optimization method of which extreme working condition is determined based on the design experience(Experiencebased Working Conditions Topology Optimization,EWC-TO).And the results showed that compared with the EWC-TO method,the maximum end deformation of the exoskeleton optimized by OEWC-TO reduced by an average of 6.78%.This ensured that the over-optimization of parts in the existing method,which would lead to the deformation of the robot end cannot meet the requirements of use,can be avoided effectively.Aiming at the problem of low accuracy of continuous movement intention recognition of human upper limb rehabilitation exoskeleton robot,a muscle contraction pattern intention recognition algorithm based on long-short-term memory(LSTM)artificial neural network is proposed,combined with motion pattern recognition based on Support Vector Machine(SVM)Algorithm to improve the accuracy of continuous movement intention recognition in various movement modes of the upper limbs.The comparison experiments between the motion pattern recognition model and the muscle contraction pattern recognition model are completed respectively,and the effectiveness of the proposed method is verified.According to the clinical characteristics and key points of rehabilitation training in the acute phase,recovery phase and sequelae phase of the upper limb rehabilitation process of hemiplegia patients,corresponding control strategies are designed.System dynamics analysis and system friction and dynamic parameter identification were completed.And,a robot joint trajectory planning method based on quintic polynomial interpolation method was proposed to improve the smoothness of the theoretical motion trajectory of the upper limb rehabilitation exoskeleton robot joints.An adaptive robust control strategy based on Udwadia-Kalaba(U-K)theory is proposed to ensure the applicability of the robot controller to model parameter changes,and also to ensure that the robot joint motion control meets the stability and smoothness requirements of trajectory tracking.Based on the upper limb rehabilitation exoskeleton robot system,the effectiveness of the motion intention recognition algorithm and the exoskeleton joint trajectory tracking control algorithm designed in this paper were experimentally verified.The experimental results of online recognition of motion intentions show that the motion pattern recognition algorithm designed in this paper has an accuracy rate of 96.11%and a system response time of 60.64±4.40 ms,the muscle contraction pattern recognition algorithm designed in this paper has an accuracy rate of 88.33%and a system response time of 444.65±43.32 ms.The joint trajectory tracking control experiment based on force/position feedback verified the accuracy and effectiveness of the adaptive robust control strategy based on the U-K theory designed in this paper.