| Forearm amputees have to face the low accuracy of prosthetic control and feedback in the adapting process of electromyography(EMG)prosthesis.With the utilization of the high-resolution information provided by high-density electromyography(HD-sEMG),we combined the prosthesis design and adaption process.By studying the natural interaction between forearm amputees and prosthesis,we proposed a new EMG prosthesis adaption process.The purpose of the study is to help amputees recover their capacity of control and living.We found the myoelectric control model of prosthesis has poor generalization and force decoding through literature analysis.The prosthetic control model has to be pre-trained the before amputees using a prosthesis,which affects prosthesis adaption efficiency.Besides,as to the prosthesis information feedback,the existing prosthesis cannot monitor rehabilitation intensity and muscle fatigue.We used independent component analysis(ICA)to extract the micro feature of HD-sEMG to realize the motion recognition and rehabilitation intensity.The hand tactile triggered by using functional electrical stimulation has been realized to solve the feedback limitation of human-machine interaction.The main work of our research includes the following:Firstly,this dissertation proposed a generated motion recognition model based on the"motor unit voting" to achieve the patients' finger motion without model pre-training.The motor unit action potential(MUAP)was extracted from HD-sEMG by ICA on the generalized motion recognition model.The proposed generalized model can identify the patient's motion.The prosthesis adaption process can be fulfilled without pre-collecting the patient's data,improving the forearm EMG prosthesis adaptation's efficiency.Secondly,in the fundamental data contribution,the dissertation provides open access"High-densitY Surface Electromyogram Recordings(Hyser)" to fill in finger force gaps in existing datasets.We have disclosed all five sub-datasets:Pattern recognition dataset,MVC dataset,One-degree of freedom(DoF)dataset,N-DoF dataset,Random dataset to solve the lack of data foundation in rehabilitation.The purpose of this work is to promote the development of rehabilitation engineering.Our work also provides basic data for researchers around the world.Thirdly,in the quantitative exercise intensity monitoring,this research compared two muscle fatigue-induce patterns.Our results showed that MUAP coherence increased significantly(p<0.05)in four frequency bands(Delta,Alpha,Beta,and Gamma)across the two contraction patterns and no significant difference(p>0.05)was observed between two contraction manners in the fatigue-induce process.Therefore,motor unit synchronization can be regarded as an indicator of exercise intensity in rehabilitation training.Fourthly,the dissertation proposed an electrical stimulation array in tactile feedback to induce the tactile sensation in the relevant hand area.The results proved our stimulus array successfully generated the tactile sensations in the hand-relevant regions.Besides,our method solved the low efficiency of nerve finding with the dual electrode.Finally,this research included a virtual prosthesis adaption service system based on the above achievements,including receive and cure module,hospital rehabilitation module,transition of hospital module,non-supervision module.The service system is designed to cover the entire rehabilitation process of prosthetic adaptation.In this dissertation,forearm myoelectric prosthesis's adaptive problems are studied from five aspects:generated motion recognition model,the Hyser dataset,muscle fatigue monitoring,tactile sensation feedback,and myoelectric prosthesis adaptation service design.The technique solves the limitation of forearm myoelectric prosthesis adaption.The integration of service design promotes the integration of "control" and "feedback "in the forearm EMG prosthesis to improve the efficiency of patients' adaptation and help them obtain better prosthetic control ability and living ability. |
PDF Full Text Request