Study on Coal Body Structure Prediction Method Based on Machine Learning and Multi-Attribute Seismic Data Integration

The structure of coal bodies is the product of brittle or ductile deformation in coal reservoirs under tectonic stress, serving as a crucial parameter influencing pore distribution characteristics, permeability, adsorption-desorption capacity, and mine safety in coal reservoirs. It holds significant research importance for the exploration and development of coal resources and coalbed methane (CBM). Under stratigraphic temperature and pressure conditions, the chemical structure and physical properties of coal reservoirs undergo corresponding deformation and evolution, leading to changes in the stress field surrounding the coal reservoir, as well as alterations in coal rock strength, pore characteristics, adsorption-desorption capacity, and permeability. Seismic data encompasses various attributes such as amplitude, frequency, and phase, with distinct differences in rock physics attributes among different coal body structures, which are closely related to seismic attributes. Through multi-attribute analysis, seismic attributes associated with coal body structures can be extracted. Machine learning is capable of processing and interpreting the nonlinear relationships between vast amounts of seismic data and rock physics attributes. By establishing a coal body structure prediction model based on machine learning technology, the accuracy of coal body structure predictions can be enhanced, allowing for an understanding of the distribution characteristics of tectonic coal in the study area and providing a reference for CBM (methane) extraction, thereby effectively improving the efficiency of CBM (methane) mining.