Estimation of Geophysical Properties of Sandstone Reservoir Based on Hybrid Dimensionality Reduction with Elman Neural Networks

Because of the complex processes and the high cost of rock geophysical properties tested in laboratory, an intelligent method based on hybrid dimensionality reduction with Elman neural networks was proposed to estimate the geophysical parameters of sandstone. Firstly, the grey correlation analysis is used to calculate the correlation between rock slice feature parameters and geophysical properties to select some parameters with high correlation; secondly, the principal component analysis is used for once more dimension reduction based on the selected feature parameters; finally, the Elman neural networks is applied to find the mapping relationship between rock slice feature parameters and geophysical properties within it. The result showed that the average relative errors of porosity and permeability were 7.28% and 6.25% respectively.