Nolinearity Analysis And Its Application On EMG Signals

The Higher-Order Spectra ( HOS )techniques can reveal more information about non-Gaussian and non-linearities than conventional techniques. This advantage may be useful in electromyographic(EMG) signals processing, and the application of HOS techniques in EMG analysis will possibly provide a new powerful way for EMG classification.This paper proposed a novel method for detection of biomedical signals based on the parametric model and higher order spectra technique. The new method can separate continuous signals automatically and this method is also useful in nonstationary time-varying signals. We proposed a sequential procedure that does not use the information provided by the future and the computational complexity is less than global segmentation of the process. This is especially useful in on-line processing. This method can reflect local signal feature and well perform in the experiments. We also present an integrated electromyographic signal (EMG) pattern recognition scheme. The application of an artificial neural network(ANN) technique together with a feature extraction technique, for the classification of EMG signals is described. We use higher order spectra technique to extract the features of EMG signals and classify them with a three-layered feed-forward network which implements the back propagation of error learning algorithm. This method has a great potential in practical application of pattern recognition.