The paper is devoted to modeling the time-frequency structure of a given class of geoacoustic impulses. A combination of time-frequency, sparse approximation, statistical and intelligent analysis methods is used for modelling. The signal time-frequency representation is constructed using the modified algorithm of adaptive matching pursuit. It is shown that the constructed representations are divided into classes by hierarchical clustering methods. Four classes of geoacoustic impulses have been obtained and analyzed. We suggest revealing internal patterns of the structure of a given impulse class using the algorithms of mining association rules and statistical analysis. Then we used the algorithms in order to model geoacoustic emission impulses. The developed algorithms for analysis and modeling of geoacoustic impulses are implemented using the MATLAB programming language. The designed system has been tested on typical geoacoustic impulses of characteristic shape. The correspondence of real and model impulses is shown, which proves the correctness of the internal structure patterns.

Authors: O. O. Lukovenkova, A. B. Tristanov, V. V. Geppener

Direction: Informatics and Computer Technologies

Keywords: Geoacoustic emission, geoacoustic impulse, time-frequency structure, intelligent analysis of signals, sparse approximation

View full article