A method for automatic detection of seismic events based on the analysis of singular entropy of a time signal

Abstract

The invention relates to seismology and digital signal processing and can be used for the automatic detection of seismic events based on the analysis of longitudinal P-waves. The proposed method is based on the analysis of structural changes in the seismic signal using singular entropy. The seismic signal is converted into digital form, divided into time segments, and a multidimensional representation of the signal is generated. For each section, a singular value decomposition is performed, which is used to calculate the singular entropy. The entropy change over time is then analyzed, including its absolute values ​​and rate of change. The occurrence of a seismic event is determined by exceeding specified threshold entropy values ​​or by its change, which corresponds to the occurrence of a structural transition in the signal associated with the arrival of a P-wave. The technical result is an increase in the accuracy and stability of detection of seismic events under conditions of a low signal-to-noise ratio, a reduction in computational complexity and the possibility of implementing the method in real time without the use of trained models.