Estimation of torque for knee joint using frequency domain features for rehabilitation robot biomechanics

Electromyography (EMG) signal includes information of intended torque, angle and impedance parameters of the knee joint dynamics. In this paper, Generalized Regression Neural Network (GRNN) and Extreme Learning Machine (ELM) have been used in trying to estimate torque and corresponding angle about knee joint. Mean Frequency (MNF), Median Frequency (MDF), Mean Power (MNP), Total Power (TTP), Power Spectral Density (PSR), Spectral Moment (SM), Power Spectral Ratio (PSR) are few frequency domain features extracted from the Fourier Transform of the EMG signal that are acquired at different angles about knee joint during extension.