M.Sc. Tezi Görüntüleme

      There is an ever-growing demand for prosthetic limbs as active electrodes, high performance processors and high-energy density batteries are used to make wide variety of limbs for disabled people. One of the most significant limbs is prosthetic hand since patients need complex hand movements to manipulate objects. But, it is not easy to design prosthetic hand because of large number of muscles and joints enabling a high degree of freedom.

      Acquisition of Electromyographic (EMG) signals and classification of patterns, the first two steps in building successful prosthetic hand controlled by complex tags from the processor running classification algorithms, are emphasized in this thesis. Surface EMG signals of forearm flexor and extensor muscles of voluntary participants, emerged as a result of 7 basic hand movements are acquired by active electrodes, amplified, digitized and then transferred to a computer for processing. EMG signals are filtered first and then pre-processed before time and time-frequency domain features are extracted. Combinations of time and time-frequency domain features are then used to train an artificial neural network (ANN) to perform classification. Correct classification ratios for both data from the training set and test set which the ANN is not trained for are reported besides significant findings.

      Key Words: EMG Signals, Myoelectric Control Systems, Electromyography, EMG features, Artificial Limbs, Artificial Neural Networks.