Processing and classification of emg signals

Lots of researchers try to modelling perfect body motions since electrochemical basis biologic signals have been discovered. In many studies, raw Mioelectrical Signals (MES) which are consequent of muscle contraction become meaningful and useful via new technics and methods. This signals used for source signals for electronic devices drive mechanical parts of prosthesis-bionic limb. Also the development of technology in robotics field; researchers wants to bring robots in hand proficiency which are rarely used in automotive, surgery etc. areas.Nowadays prosthesis hands used has limited activity ability. To compose more proficient prothesis-artificial hands the studies must keep going on this three topics. The first one is mechanical solutions which supply the ability of freenes scale, second one is electronic circuits which are responsible for obtaning motion speed at desirable ability and the third one is to generate source signals which are used to drive this electronic circuits. The third one become better which is paralel with the improvement and development of signal processing and artificial intelligence technics. We couldn?t talk about more developed prothesis-artificial limbs unless third and second problems solved.In this work, the aim is to make better and newer solutions for third problem. For this reason, motions which are used for controlling objects consciously are identified and Electromyogram signals which are used for talented some hand motions was recorded according to this consciously motions. After this recording stage; Wavelet Transform based autoregressive models that is suitable for signal nature and known as a good signal processing technic for Electromyogram used for analysis of signal. After that these signals are classified by neural networks.