Design of Electromyography classification system using Artificial Neural Network

In this paper we report on a system design for automatic classification of surface Electromyography (EMG) signals using Artificial Neural Network as a classifier. Key requirements to the system components are shortly described together with the main features and challenges in the field. The system comprise of wireless measurement system to measure, record and transfer EMG signal to signal processing and classification unit (SPU). The SPU implements digital filter and a feed-forward neural network for the classification of EMG signals related to muscle contraction and relaxation. The functionality of the designed and realized system is demonstrated on a classification of the set of real EMG signals measured on proband.