Klasifikasi Gerakan Pergelangan Tangan Berbasis Sinyal Elektromiografi dengan Fitur Root Mean Square dan Pengklasifikasi Support Vector Machine

Abstract

The wrist is essential to various human activities, with basic flexion, extension, and straight (normal) movements. These movements can be identified using surface electromyography (sEMG) signals that allow human interaction with computers through muscle activity, referred to as a muscle-computer interface. This research aims to perform the classification of flexion, extension, and normal movements as an initial stage of developing a muscle-computer interface system. The main stages of this research include data acquisition, signal processing, and motion classification with a support vector machine (SVM) offline and not in real-time, using root mean square (RMS) features with a moving window. This study successfully designed an effective sEMG signal measurement and processing method for motion classification. The designed classification algorithm showed high performance with an accuracy of 90.27%, precision of 90.61%, sensitivity of 90.17%, and f1-score of 90.27%, demonstrating the ability of the muscle-computer interface to classify wrist movements with a high degree of accuracy.