Research and application of Intelligent high resolution processing method based on ISTA-Net
Huahui Zeng, Qin Su, Sanyi Yuan, Lide Wang, Xu Yanwu, Huijie Meng, Deying Wang · 2024
With the deepening of seismic exploration and development, the target of seismic exploration has gradually changed to deep and ultra-deep oil and gas exploration. In recent years, the Permian Maokou Formation dolomite reservoir in Sichuan Basin of China has made a major breakthrough. However, due to the deep burial, serious absorption attenuation, narrow bandwidth and low dominant frequency, and the thin thickness and strong heterogeneity of dolomite reservoir, high-resolution imaging and fine prediction of thin reservoir are difficult, so it is urgent to study the high-resolution processing method of deep thin reservoir. In view of the limited frequency band of seismic data, unknown seismic wavelet and high frequency attenuation of seismic waves in the propagation process, the traditional high-resolution processing method is not suitable. This paper studies ISTA-Net method, builds a network based on the iterative process of optimization algorithm, accurately represents the mapping relationship between input data and tags, and reduces the dependence on training samples. The actual data test shows that the proposed intelligent high-resolution processing method based on ISTA-Net improves the resolution processing accuracy of thin deep carbonate reservoir, which can be popularized and applied to high-resolution imaging and high-precision prediction of thin reservoirs.