Comparative Analysis of Neural Network Techniques for Upper Limb Movements

Mehak Saini and Himanshi Saini

Electromyography (EMG) is used to record muscle activation patterns while performing functional activities. In this paper, a comparative analysis of different neural network techniques has been performed for the classification of upper limb movements. The EMG signals are acquired using Biopac MP 150 systems from six muscles of upper extremity while performing four different locomotion activities. Three features calculated for the classification of four arm movements are median frequency and total power. Five different Artificial Neural Network (ANN) techniques have been implemented for the classification purpose. A comparison of these techniques shows that probabilistic radial basis function network is fastest and gives 100 % accuracy for the classification of the movements. These results can help in improving the quality and design of prosthetic upper limb.

Volume 11 | 09-Special Issue

Pages: 81-86

DOI: 10.5373/JARDCS/V11/20192539