Automatic segmentation of seismic signal with support of innovative filtering

Abstract

The paper presents a new algorithm for automatic segmentation of seismic signal. This segmentation relies on the observation of changes in the signal spectrum structure based exclusively on second-order signals statistics. The starting point of the proposed method are time-varying reflection coefficients of adaptive innovation filter. Schur's algorithm, performing the estimation of the reflection coefficients, is known as a fast, accurate and stable algorithm capable of quickly tracking the changes in signal statistics. Signal segmentation is a necessary stage for further processing of the signals, their interpretation, modelling or automatic classification. The issue of automatic segmentation is known in literature for its many applications (diagnostics of machinery, speech analysis, medicine etc.). There have also been a few proposals for segmentation of seismic signals. Unfortunately, segmentation of seismic signals is very difficult because of their variability. The purpose of the authors' work was to design an algorithm, which could be used in automatic processing of seismic signals in a deep underground copper mine. The paper presents the algorithm structure, defines its properties, specifies the values of parameters and presents the results of the effectiveness of the detection process. The obtained segmentation results are satisfactory.