Anomaly Detection in Pulse Geoacoustic Signals

• DOI

The article proposes a method for describing pulse geoacoustic signals, which allows us to detect anomalies. To do this, the geoacoustic signal is converted into a symbolic sequence using the structural-linguistic analysis method. Based on this sequence, an alphabet of the analyzed signal consisting of unique symbols is compiled. Each symbol is described by a binary square image matrix. The frequency of each alphabetic symbol is calculated for a given period. The distributions of pulses over alphabetic symbols obtained for successive observation periods of a given duration can be represented graphically. The horizontal axis corresponds to the image matrix order. The vertical axis corresponds to time. The symbol frequency is indicated by color. Such graphical representation allows us to detect anomalies associated with pulse waveform and repetition period changes. The method was tested on artificial and natural data. The proposed method was compared with the spectrogram method.

Gapeev M., Lukovenkova O., Senkevich Yu., Solodchuk A. Anomaly Detection in Pulse Geoacoustic Signals // 2022 VIII International Conference on Information Technology and Nanotechnology (ITNT), 2022, pp. 1-4, doi: 10.1109/ITNT55410.2022.9848593.