Application Research Of Surface EMG Signal In Upper Limb Rehabilitation Training

In recent years,the body movement disorders caused by stroke and brain trauma have increased,and the reconstruction of motor function has attracted widespread attention.Most of the current rehabilitation training has the disadvantage that the patient’s active participation is not high enough.In response to this problem,this paper studies the sEMG signal of the upper limb in rehabilitation training,and the sEMG signal is combined with the rehabilitation training platform to verify the feasibility of applying the sEMG signal to rehabilitation training.The sEMG signal contains a large amount of limb movement information.By analyzing and processing the sEMG signal,the human muscle movement state is grasped,and the movement intention of human upper limbs is identified.This article uses the DELSYS wireless surface myoelectric acquisition device to collect myoelectric signals of human upper limbs.Based on the upper limb rehabilitation robot platform of the project team,four movements frequently involved in the rehabilitation process of patients with upper limb dysfunction were selected as research objects,namely elbow flexion,elbow extension,shoulder adduction and shoulder abduction.The work of this paper is mainly divided into four parts: sEMG signal acquisition,preprocessing and feature extraction,signal recognition classification and its application in rehabilitation training platform.1.Acquisition of myoelectric signals.The DELSYS wireless surface electromyography acquisition system was used to acquire the myoelectric signals of the subject’s upper limb movements.the electromyographic signal data was converted into CSV table form by data transformation unit according to categories,so as to realize the electromyographic signal processing by subsequent Matlab programming.2.Pretreatment and feature extraction of myoelectric signals.The sEMG signal is denoised in two ways: filter and notch noise reduction method and wavelet denoising method.The noise reduction effect is evaluated by the SNR,and finally all the EMG signals are processed by the wavelet denoising method.The feature extraction is performed in the time domain and the frequency domain,and finally the root mean square,the absolute value mean,the median frequency,and the average power frequency constituting the feature vector is input into the classifier for pattern recognition.3.Identification and classification of myoelectric signals.The vector machine is used for classification and the parameters in the support vector machine are optimized by the grid search method and the particle swarm optimization algorithm.The test samples are classified using atrained support vector machine.Compared with the classification results of support vector machines optimized by grid search,the PSO-optimized vector machine recognition rate is 7%higher,reaching 85%.4.Combining the pattern recognition algorithm of myoelectric signals with the upper limb rehabilitation training platform to verify the feasibility of using EMG signals for rehabilitation training.The upper computer sends a motion command to the controller according to the classification result of the support vector machine to make the rehabilitation training platform move.The experimental results show that the average correct implementation rate of the four training actions of the rehabilitation training platform is 80%.The motion recognition method for the upper limb motion is obtained by analyzing and processing the surface myoelectric signal.Combining the surface EMG signal with the upper limb rehabilitation training platform to verify the possibility of using the upper limb EMG signal on the healthy side to drive the affected side for rehabilitation training.The experimental results show that the average correct implementation rate of the four training actions of the rehabilitation training platform is 80%.