| Scientific and comprehensive rehabilitation exercise is of great significance in accelerating the remodeling of neuromuscular motor function in stroke patients and improving their motor function.With the development of rehabilitation medicine theory and the new rehabilitation techniques,virtual reality technology has gradually been applied to rehabilitation training with its powerful interactivity,immersion and conceivability.In view of the limitations of the existing virtual rehabilitation system in terms of safety,individual adaptability,and effectiveness of rehabilitation evaluation,this paper studies virtual rehabilitation technology with biological information feedback,by adjusting parameters of virtual scenes to avoid factors such as improper training intensity and secondary injury caused by fatigue in rehabilitation exercises.At the same time,the directional synchronization coupling relationship between brain myoelectrics in virtual rehabilitation training was analyzed,and the differences in neurophysiological mechanisms between healthy people and patients were explored from the view of neuromuscular movement control,which will be further applied to the evaluation of motor function in virtual rehabilitation system as theoretical basis.Firstly,the research status of rehabilitation training technology was summarized,the significance of biological information feedback based virtual rehabilitation was explained.The mechanism characteristics of EEG and EMG signals,as well as the acquisition method of EEG and EMG based on this system were introduced.The corresponding pretreatment method was used to remove the noise from measured EEG and EMG signals.The characteristic index and classification method as well as features of brain fatigue and muscle fatigue were determined to recognize motion paradiagm in virtual training,which would provide a basis for motion recognition and fatigue estimation in virtual rehabilitation.Secondly,in order to analyze the directional and synchronous coupling relations between EEG and EMG during virtual rehabilitation training,the coherence analysis and Granger causality analysis method were analyzed and compared.A variational modal decomposition-partial directed coherence analysis method was proposed to quantify the directional synchronization relationship of EEG and EMG in their natural frequency band,which would provide basis for evaluation of rehabilitation state in training.Thirdly,aiming at the research needs about individual adaptability and safety of virtual rehabilitation system,EMG feedback was applied to human-machine interaction,and the corresponding actions in virtual scenes are completed based on the pattern recognition results of EMG.On the other hand,a firefly algorithm-fuzzy neural network algorithm was proposed to adaptively optimized the width of the fuzzy neural network membership function and the network weight.Furthermore,based on this method,a virtual scene parameter adjustment strategy based on the firefly fuzzy neural network was designed,which can quickly and accurately map the mapping from the input brain electromyography to the output control parameter set.Finally,the methods proposed in this paper were verified based on the virtual rehabilitation system.Based on the virtual rehabilitation training system,four healthy individuals and four patients were selected to complete the test of motion pattern and scene motion recognition rate,the adaptive adjustment test of the system scene motion control coefficient,to verify the validity of the firefly algorithm-fuzzy neural network parameter control strategy for virtual scenes.At the same time,the directional coupling relationship between brain and muscle in rehabilitation exercise was quantitatively described based on variational modal decomposition-partial directed coherence analysis method proposed in this paper,to evaluate the state of rehabilitation training,which would provide the theoretical basis for objectively motor function assessing during virtual rehabilitation. |
PDF Full Text Request