EMG-muscle force estimation model based on back-propagation neural network

Electromyography (EMG) signal precede many information about muscle activity, which can be used to analyze muscle functions, such as muscle forces. Muscle forces estimation is important for orthopedists, bio-mechanists, and physical therapists because of joint contact forces. In this work, EMG signal is used to estimate human arm muscle force by implementing a new Back-Propagation neural network (BPNN) model. The proposed model, uses the rectified smoothed electromyography (RSEMG) signals as input to generate the estimated muscle force as an output. To train the system, the Biceps-brachii muscle force was used. To validate our model, the Brachioradialis, Triceps-brachii, and Flexor carpi ulnaris muscles´ signals were used. To measure the performance of the model, the BPNN output estimated force is compared with the Hill-Type force estimation. Our results illustrate that. The output of our model is accurate compared with Hell-type model. This can be proved through the regression value of our model which exceeded 99%. We have used four muscles of the arm to show that our model can estimate the force of any muscle of the hand, which makes it a global hand muscle force estimator.