The application of FastICA algorithm based on wavelet domain to seismic data denoising of unconventional oil and gas reservoirs

In recent years, with the continuous expansion and development of Sulige gas field in Ordos Basin, the digital 3D seismic data have been widely used, and digital geophone with high sensitivity that causes noises in seismic data is extremely developed. Therefore, the digital 3D seismic data prestack denoising is the key to data processing. In this paper, based on the 3D digital seismic data, in combination with the new development of the blind signal separation technology, especially the latest development and application of the neural network technology, the authors studied and established the blind signal separation algorithm model suitable for theoretical seismograms, performed blind source separation technology for known seismic reflection data, transformed the seismic signal into wavelet domain, used FastICA algorithm to conduct blind separation denoising, and then transformed the denoised signal from the wavelet domain to the time domain. The experimental results show that the denoising result of this method is better than the direct denoising result in the time domain.