Evaluation of Muscle Fatigue based on SEMG using Deep Learning Techniques

This research study analyzes the problem of identifying muscle exhaustion using complicated and unstable sEMG data. In this study, three different approaches are used to evaluate muscular tiredness. The Feature Fusion Network (FFNET), a revolutionary strategy, is what the researchers suggest as the initial technique. To increase the precision of muscle fatigue diagnosis, the FFNET combines one-dimensional feature extraction techniques like Root Mean Square (RMS), Average Amplitude Change (AAC), and Mean Absolute Value (MAV) with two-dimensional feature extraction techniques like Fast Fourier Transform (FFT). The sEMG signals are used by the FFNET model to extract and concatenate features. Long Short-Term Memory (LSTM) and machine learning classifiers are then supplied with these characteristics. In the second approach, classification is handled by LSTM while feature extraction is handled by Convolutional Neural Network (CNN) layers. CNN does the assessment of muscular exhaustion in the third technique. Accuracy, recall, precision, and F1 score are computed for each of the three techniques stated above and compared in order to assess muscle fatigue. The outcomes show that the suggested FFNET model beats previous approaches in terms of accuracy, achieving an accuracy of 99 percent in LS TM.