Real-time Monitoring Algorithm of Muscle State Based on sEMG Signal

Abstract

Abstract Muscle fatigue can occur when we do too much physical activity in a short period of time, or if we have not been exercising regularly. For patients who need rehabilitation training, muscle fatigue has certain safety risks. Therefore, real-time monitoring of muscle fatigue is very necessary and meaningful. At present, there are few studies on the assessment of muscle status, and most of the methods are limited to the detection of fatigue threshold, lagging behind the occurrence of fatigue. This paper analyzes the change law of the fatigue process model and summarizes it, and designs it into a real-time evaluation algorithm of muscle state. First, the autoregressive moving average (ARMA) model is constructed utilizing surface electromyography (sEMG) and joint motion angle data. Next, the parameters are identified and monitored in real-time. Finally, hierarchical clustering is employed to categorize levels of fatigue. The algorithm is capable of capturing changes in fatigue-induced correlations between sEMG and the corresponding joint motion angles during dynamic movements, enabling accurate classification of muscle fatigue states. Experimental results have demonstrated the effectiveness of the algorithm, which has significant implications for real-time evaluation of human muscle state and the design of control systems in human-computer interaction.