Automatic Muscle Fatigue and Movement Recognition Based on sEMG Signals

This paper explores the potential of wearable technology equipped with surface electromyography (sEMG) sensors to provide real-time monitoring of muscle activity during dynamic contractions. By leveraging sEMG data, the study aims to classify forearm exercises and detect muscle fatigue, contributing to a deeper understanding of physical performance and well-being. Through the implementation of a cost-effective sensor armband in a laboratory setting, the feasibility of using sEMG for quantifying muscle fatigue and recognizing forearm exercises is demonstrated. The acquired sEMG signals are analyzed using a set of features in both time and frequency domains. Indicators such as Root Mean Square (RMS) and Mean Frequency (M N F) are utilized to assess fatigue levels. Classification of signals into fatigue and non-fatigue categories is achieved through linear regression slope analysis, followed by a more nuanced classification into three exercise categories using neural network architecture. We evaluate the proposed methods using an original database of 100 subjects, accessible for free through the provided link. Support Vector Machines (SVM) yield the best results, with a classification accuracy of 74% for both exercise and muscle fatigue.