Research of Electromyographic Signals Processing Methods

All bionic prosthetic control systems are based on the analysis of electrical signals traveling from the brain to the muscles of an amputated limb, their processing and transformation into commands for a bionic device. One of the most effective methods is direct myoelectric control and recognition of myoelectric motion models with invasive and noninvasive sensors based on the same principle of receiving a signal from muscles – electromyography. To register an electromyographic signal, various types of electrodes are used, the choice of which depends on the objectives of the study and the depth of the location of the muscles under study. In this work, the method of bipolar noninvasive recording was used, which allows tracking the total activity of a particular muscle. Registration of biopotentials is complicated by the presence of various interferences, which requires the use of various methods of preprocessing and filtering signals. To get rid of the induced voltage, a notch filter was used, and high- and low-pass filters with adjustable cutoff frequencies were used to limit the signal in frequency. Digital signal processing is a key component in the analysis and interpretation of electromyographic signals. It can be used to automatically segment these signals and detect events such as the beginning and the end of muscle contractions. In particular, such filters with finite impulse response as a moving average filter and a median filter with a configurable filtering window were considered. When considering the methods of preprocessing and digital filtering of electromyographic signals, it can be concluded that the use of combined processing methods provides the best results.

Authors: R. O. Semchugova, A. A. Uhov, D. K. Kostrin

Direction: Physics

Keywords: bionic prosthesis, electromyography, muscle activity, signal filtering, notch filter, digital processing, moving average filter, median filter


View full article