Muscle Fatigue Analysis for Healthy Adults Using TVAR Model with Instantaneous Frequency Estimation

The objective of this paper is to design a nonstationary time-varying autoregressive (TVAR) cascaded model to analyze electromyography (EMG) signals by using instantaneous frequency for muscle fatigue assessment. EMG is commonly used in the muscle fatigue study during muscle contractions by analyzing myoelectric signal spectrum. To validate the findings, our results are compared with the conventional short time Fourier transform (STFT) method. STFT has limitations in joint time frequency resolution for long intervals, whereas TVAR models overcome these limitations for nonstationary signals. In this study, EMG data recorded from the rectus femoris muscle are used to characterize muscular fatigue. Characterizations are done by using mean frequencies (MNF). According to our results, the new method has a better accuracy in signal representation, frequency resolution and joint time distribution