Electronic control arm using electromyographic signals

The studies focused in pattern extractions of electromyography signals (SEMG) has been growing, due to their multiple applications. This paper presents an electronic system implementation for the SEMG recording of a subject upper extremity in order to remotely control an electronic arm. Initially, we performed a signals preprocessing, to remove the less important information and to recognize the interest areas. Then the patterns were extracted and classified. The techniques used were: The wavelet analysis (AW), the principal components analysis (PCA), the Fourier transformed (FT), the discrete cosine transformed (DCT), the support vector machines (SVM) and the artificial neural networks (ANR). In this paper we demonstrated, that the methodology stated, allows to realize a process of classification with a superior performance to 95%. There were recorded more than four thousands signals.