The article discusses the use of algorithms based on artificial neural networks when working with non-stationary signals, in particular biomedical ones, such as an electroencephalograph signal, to identify and process local signal features. The use of conventional patch electrodes for EEG recording leads to the appearance of noise and requires special signal processing. For this, bandpass wavelet filtering is used. The obtained data are further processed using an artificial neural network to identify information contained in a limited interval in the biomedical signal. To train the neural network, the Levenberg-Marquardt method was used, as the optimal one and meeting the requirements.
Keywords: artificial neural network, electroencephalography, wavelet filtration, biomedicine, non-stationary signal, system analysis