Design Of Forearm Fatigue Evaluation System Based On SEMG Signal

Stroke disease is very easy to recur,and has a high mortality rate and a very high disability rate.According to epidemiological calculations,of the 75% of survivors,more than 80% of patients have obvious disabilities after natural recovery.Among these disabled patients,85% have left upper limb motor dysfunction.After 3-6 months of rehabilitation treatment,only about 30% of the patients have restored upper limb motor function.Therefore,upper limb rehabilitation training is particularly important.The detection of muscle fatigue in upper limb rehabilitation training can effectively avoid the secondary injury caused by overtraining during the rehabilitation process.This article takes young normal people as an example,divides the degree of fatigue according to the RPE table,and evaluates the forearm fatigue with different degrees of fatigue of the upper limbs after the elbow and wrist extension and flexion exercises.The writing of this article is arranged according to the experimental process,which includes surface EMG signal acquisition,effective segment selection,EMG signal preprocessing,feature extraction,and judgment of fatigue.The instrument for collecting EMG signals in this article is EMG SystemTB0810;we require the tester to complete the elbow extension and flexion,and the wrist extension and flexion.The main completed actions are to keep the upper arm stationary and start the 1.25 kg barbell from the body side as 0 ° Lift to 135 °,pause for 5 seconds,and then return to the body side;keep the arm still,start to flex the wrist at the maximum extension of 0 ° with a weight of 1.25 kg,and then flex the wrist at 75 °.The TKE operator and the effective part of the sEMG signal are detected based on short-term energy,and three types of common noises in the signal(power frequency noise,baseline drift,and white noise)are de-noised and pre-processed;the spectral interpolation method,The mathematical morphological theory and the EMD threshold are used to eliminate the noise and verify these methods.Four main characteristics of mean power frequency(MPF),median frequency(MF)integrated electromyogram(IEMG)and root mean square(RMS)were selected to analyze the degree of fatigue of men and women.Based on the BP neural network,the paper evaluates the fatigue of the collected samples;and proposes an optimization scheme for the shortcomings of the gradient descent neural network algorithm,which is slow in convergence and large in limitations.Finally,the gradient descent neural network and Bayesian normalization are compared.BP neural network fatigue classification in two ways.