Three-dimensional seismic data waveform semi-supervised clustering method based on EM algorithm

Abstract

The invention provides a three-dimensional seismic data waveform semi-supervised clustering method based on the EM algorithm. According to the three-dimensional seismic data waveform semi-supervised clustering method based on the EM algorithm, following processing is conducted on three-dimensional seismic data in a time window on a target layer; the extreme point of a three-dimensional seismic data waveform is searched, a seismic waveform is fit through the Chebyshev polynomials, and a fitting coefficient is taken as a waveform characteristic parameter; fit seismic waveforms of well byway seismic data in the three-dimensional seismic data is classified according to logging information, so that a labeled sample data set containing class information is formed; semi-supervised clustering is conducted on fit seismic waveforms which are not classified according to the EM algorithm, wherein the parameter initial value for iteration of the EM algorithm is given through the labeled sample data set containing the class information, and then clustering is conducted on the waveform characteristic parameter according to the fact that waveforms around the extreme points on the same geologic horizon are similar. According to the three-dimensional seismic data waveform semi-supervised clustering method based on the EM algorithm, logging data are adopted during clustering, classifying precision is improved, and a classification result and actual class information are closely associated.

Description

technical field [0001] The invention relates to the technical field of seismic exploration data interpretation, and more specifically, relates to a method for classifying waveforms of three-dimensional seismic data used in petroleum seismic exploration, geological structure analysis, and the like. Background technique [0002] Waveform classification technology based on seismic signals is an important means for seismic interpreters to analyze underground reservoirs and stratigraphic structures. Reasonable and accurate seismic signal waveform classification results can truly reflect the underground reservoir and stratigraphic structure, which is conducive to seismic interpreters to accurately interpret the underground structure, thereby improving the prediction of lithology, sand body prediction, and fractured oil and gas. Reliability of reservoir prediction and hidden oil and gas reservoir prediction, thereby reducing exploration risks, saving exploration costs, and bringing...

AI Technical Summary

Problems solved by technology

However, using different clustering methods and distance calculation methods, the classification results obtained are often different, which affects the classification accuracy, and the unsupervised classification method ignores some important prior information and is not closely related to the actual category information.

Images