Research On Control System Of Hand Rehabilitation Robot Based On SMA

Stroke is a common acute cerebrovascular disease with high morbidity and incidence trend at a younger age in recent years.Stroke has enormous hazards,leading to the common sequelae of hemiplegia of the hand,aphasia,paralysis of the lower limbs.For stroke patients with hemiplegia of the hand,their hand muscles present a clonic state at the beginning of onset.Failing rehabilitation treatment in time,patients will be afflicted with lifelong injury to the palm.For rehabilitation physicians who treat hemiplegia,the daily grind of a single rehabilitation task can wear down their patience and gradually lose the pleasure from their work.The hand rehabilitation robot was invented to address the problems and soon developed rapidly.This thesis focuses on the hand rehabilitation robot based on shape memory alloy wire,mainly including the following parts:(1)The common shape memory properties,self-sensing properties and superelasticity of shape memory alloy wires are introduced.The change of shape memory alloy wire temperature and displacement under periodic voltage was analyzed,and the feasibility of using shape memory alloy wire as a rehabilitation hand driver was verified.(2)Aiming at the control of shape memory alloy wire rehabilitation hand,two model-free control methods PID and fuzzy PID were adopted,and actual control was carried out on the rehabilitation hand system to verify the effectiveness of the selected control method.(3)Aiming at the problem of hand movement recognition of patients,SVM algorithm was used to conduct experimental analysis on Ninapro DB5 subject electromyography data set,and the results were compared with the classification results of other machine learning methods,suggesting that the SVM algorithm had high motion recognition accuracy in this data set.Secondly,a method combining sEMG data with acceleration data was proposed,namely,features of both were extracted and fused into new features,and classification and identification were conducted by the SVM algorithm.Compared with the classification results only using sEMG data,the combination greatly improved the accuracy.(4)The rehabilitation hardware platform was built,the software control system was designed,and different rehabilitation strategies were formulated according to patients’conditions in different rehabilitation periods.Patients with early disease hand muscle stiffness were unable to stretch,so the use of passive rehabilitation could assist patients in rehabilitation training.In the middle stage of rehabilitation,the patients had certain motor ability and used mirror image for active rehabilitation.At the later stage of rehabilitation,the patients’ hand motor ability was significantly improved and active rehabilitation strategy was adopted.In addition,they are encouraged to participate in the rehabilitation game to improve their willingness for rehabilitation treatment.