Forecasting the Semg Signal Using Wavelet Transform and Anfis Model

Abstract

In this paper we study how the muscles in the human body move, electromyography (EMG Signal) is employed as a diagnostic technique for identifying various muscular activity. Noise from the SEMG signal is effectively minimized with a suitable wavelet selection. The root mean square values have been evaluated to determine which wavelet is the most efficient for signal denoising. Further, since a learning method of a neural structure with connections based on rules is necessary to be able to estimate the relationship, this paper also aims to analyse an approach that uses signals obtained by surface electrodes to characterize hand movements of the human arm for pattern recognition (i.e. ANFIS method is employed). The characteristics of seven hand gestures are categorized using the ANFIS-based learning, which is then assessed in order to predict the link between input and output.