Seismic multi-attributes inversion using neural network and its application in predicting lithology of coal seam's roof

The lithology of coalbed's roof has huge impact on the safe mining. Through training the data of seismic attributes(including P-impedance) and gamma ray logging data by neural network algorithm, we can get the nonlinear relationship of them, then we apply the nonlinear relationship to the whole seismic data volume and get pseudo gamma ray data volume. Compared to P-impedance, the gamma ray data can better distinguish the sandstone and mudstone, therefore, we can predict the lithology of coalbed's roof more directly and accurately, also improve the resolution on lithology prediction issues.