Research On The Multi-mode Assist-as-needed Control Of Upper Limb Rehabilitation Exoskeleton

With the background of the increasingly serious aging problem in China and the increasing number of hemiplegic patients caused by cerebrovascular disease,the use of upper limb rehabilitation exoskeletons to replace physicians in assisting the rehabilitation training of hemiplegic patients can effectively alleviate the huge pressure faced by the social medical system.As a programmable intelligent electromechanical device,the upper limb rehabilitation exoskeleton provides repetitive and high-intensity assistive rehabilitation training for patients with different motor abilities or in different stages of rehabilitation(?accid,spastic and convalescent stages)to provide continuous,e?cient and targeted auxiliary rehabilitation training(passive,active and resistive training),and reconstructs the patient’s motor ability.However,for the system of upper limb rehabilitation exoskeleton with high degree of human-machine coupling,the dynamic interaction between the human body and the exoskeleton has strong nonlinearity and uncertainty,the accuracy and robustness of trajectory tracking during exoskeleton movement are crucial.In addition,for hemiplegic patients in different rehabilitation stages or with different motor abilities,their clinical characteristics are different,and the control method with a single assistance rehabilitation training mode can not meet the training needs of different patients.Therefore,the multi-mode assistance control strategy of upper limb rehabilitation exoskeletons,which is suitable for different hemiplegic patients and has stronger functional universality in practical application.This paper focuses on the key issues involved in the control methods of the upper limb rehabilitation exoskeleton,focusing on the assistance control method of the upper limb rehabilitation exoskeleton for the rehabilitation training of hemiplegic patients in the ?accid,spastic and convalescent stages.The main research contents include the following:(1)By analyzing the engineering feature mechanism of human upper limbs,including distribution of degrees of freedom,joint activities and limb length range,etc.,and using SolidWorks 3D design software to design a seven-degree-of-freedom upper limb rehabilitation exoskeleton virtual prototype.By further re?ning the key structure of the designed prototype,an experimental prototype of upper limb rehabilitation exoskeleton for assisting rehabilitation training was designed and processed,and an experimental measurement and control platform for experimental research was built based on the Control Desk and Matlab/Simulink software and d SPACE hardware-in-loop measurement control platform.For the designed upper limb rehabilitation exoskeleton prototype,the D-H coordinate system transformation matrix was established according to its degree of freedom con?guration and size,and the corresponding kinematics and dynamics model were derived.(2)For the task trajectory tracking-based passive control method of the upper limb rehabilitation exoskeleton suitable for patients with hemiplegia in the ?accid stage,the control method with tracking error constraints was designed based on the barrier Lyapunov function(BLF)?rst.The dynamic model of the designed upper limb rehabilitation exoskeleton were calculated by SolidWorks software.To handel the modeling error and external human-computer interaction disturbance,the time delay estimation algorithm was used,and a BLF-based robust adaptive time-delay controller was designed to improve the tracking accuracy and robustness of the system.Furthermore,considering that the BLF with the ?xed constraint boundary has strict requirements on the initial position error of the system,a BLF with time-varying constraint boundary was constructed,and the neural network was used to compensate the time delay estimation error.Then,a TVBLF-based neural network time delay control algorithm was designed.Finally,in order to estimate the velocity and external integrated disturbance of the upper limb rehabilitation exoskeleton system,a neural network-based state observer was designed,and on this basis,an adaptive output feedback controller based on state observer was constructed.The stability and accurate trajectory tracking control of the exoskeleton was realized only using the position measurement information of the system.The proposed passive control method was veri?ed by simulation and experiment based on the design of the upper limb rehabilitation exoskeleton virtual prototype and experimental measurement and control platform.(3)Aiming at the active control method based on Assist-as-needed(AAN)for the upper limb rehabilitation exoskeleton suitable for patients with hemiplegia in the spastic stage.Firstly,considering the task tracking error of the patient and the intervention of the exoskeleton on human motion during the training process,a fuzzy logic strategy was introduced to adjust the impedance stiffness of the exoskeleton online adaptively,and a force/position evaluation-based adaptive assist-as-needed control algorithm was ?nally designed.Further,a multi-indicator evaluation function of human body participation based on task tracking error,human-computer interaction force and single training task completion time was constructed,and the adaptive particle swarm optimization(APSO)algorithm was introduced to optimized the impedance stiffness of the exoskeleton and the target movement velocity of the task to maximize the enthusiasm of the human body,and ?nally a multi-indicator optimization-based assist-as-needed was designed.Finally,in order to select the appropriate training task trajectory to better adapt to patients with different athletic abilities and willingness to participate actively,an energy function for evaluating human athletic ability and identifying motion intention was constructed.Then,the adaptive central pattern generator(ACPG)was used for planning training task trajectory online and adjusting impedance stiffness of the exoskeleton adaptively,and a training task planning-based assist-as-needed control algorithm was designed.Based on the experimental measurement and control platform of the upper limb rehabilitation exoskeleton,the effectiveness of the proposed control method was veri?ed.(4)Aiming at the multi-mode assistance control method of the upper limb rehabilitation exoskeleton suitable for patients with hemiplegia in the recovery period,a virtual tunnel is ?rstly constructed based on the training task trajectory to divide the task space area,and a multi-mode assistance control including resistive mode and assistive mode is designed for different task areas.Considering the patient’s motion performance in different training modes,the fuzzy logic was introduced to adaptively adjust the levels of resistive and assistive force provided by the exoskeleton,and a virtual tunnel-based multi-mode adaptive assistance control algorithm was designed.Then,for the goaloriented rehabilitation training task,the force ?eld area for resistive and assistive training is planned based on the tracking target position,and the human movement ability evaluation function based on the task tracking error and human-exoskeleton interaction force was designed.On this basis,the force ?eld area and target motion velocity were adjusted online to adapt to patients with different motion abilities and fatigue degrees.Moreover,a target-oriented multi-mode adaptive assistance control algorithm was designed.Finally,based on the upper limb rehabilitation exoskeleton virtual prototype and experimental measurement and control platform,the effectiveness of the designed multi-mode assistance control method is veri?ed by simulation and experiments.