Study On A SEMG-based Upper Limb Rehabilitation Robot Training System

Surface electromyography(sEMG)signals are potential signals collected on the muscle surface by the surface electrode during the movement of the body limb.Different actions triggered different muscle group,thus producing different electrical signals.In this paper,the sEMG signals were processed to identify the movement pattern,so as to realize the rehabilitation exercise training for the hemiplegic patients by using the uninjured upper limb to guide the injured side upper limb.In this paper,a sEMG-based signals upper limb rehabilitation robot rehabilitation training and assessment system were developed,which consisted of the sEMG signals acquisition equipment,Upper Limb Exoskeleton Rehabilitation Device(ULERD),a computer,BLDC motor and controller,etc.In this paper,a sEMG-based signals upper limb self-rehabilitation training system for hemiplegic patients were presented to help patients achieve active training.We combined pattern recognition technology and robot technology to achieve the design of the whole system.The sEMG signals from the patient's uninjured upper limb performing different motion pattern were extracted.Firstly,the sEMG signals were preprocessed,and then the feature extraction and pattern recognition classification were carried out.Eight subjects who participated in the experiment,the average accuracy of classification for offline pattern recognition was 91%.In the on-line self-rehabilitation training,when the uninjured upper limb of the patients performed some action,the result of classification using the off-line trained classified for the pattern recognition was obtained to drive the motor to control the rehabilitation robot to perform the corresponding action.The self-rehabilitation training was achieved.The highest accuracy rate of action execution for the eight subjects was 98.3% and the average accuracy rate was 92%.In this paper,a rehabilitation assessment method that muscle force as a standard was put forward to evaluate the degree of the recovery of muscle force for patients.The sEMG signals from different muscle force were extracted,and then according to clinical medical strength grading table,muscle force were divided into six levels.As sEMG signals are nonlinear,the complexity of the sEMG signals was different under different muscle force.In this paper,the fuzzy approximate entropy(fAp En)and the power of sEMG signals were extracted to create two-dimensional characteristic vector.With the directed acyclic graph support vector machine classification method,the average accuracy of classification has reached 92%.Our research provides beneficial references to the way of rehabilitation training and evaluation method for upper limb in rehabilitation medicine field,which may help to broaden the ideas of research.