Neural computing for seismic principal components analysis

Abstract

The neural network of the unsupervised generalized Hebbian algorithm (GHA) is adopted to find the principal eigenvectors of a covariance matrix in different kinds of seismograms. The theorem about the effect of adding one extra point along the direction of the eigenvector is proposed to help the interpretations that more uniform data vectors along one principal eigenvector direction can enhance the eigenvalue. Diffraction pattern, fault pattern, bright spot pattern and real seismogram are in the experiments. From analyses the principal components can show the high amplitude, polarity reversal, and low frequency wavelet in the detection of seismic anamalies and can improve seismic interpretations.