Uniaxial Compressive Strength Determination of Rocks Using X-ray Computed Tomography and Convolutional Neural Networks

The uniaxial compressive strength (UCS) is an important parameter for rock mass classification and rock engineering designs. This study proposes a novel method for predicting the UCS of rocks using X-ray computed tomography and convolutional neural networks. First, X-ray CT scanning was conducted on five mudstone specimens. The volume data characteristics of the different density compositions in rock specimens were extracted from the CT slices according to the CT thresholds. Then, the function between the cumulative CT value and the peak strength was established. The given CT image data samples of a certain rock correspond to the scope of the predictive UCS. To approve the accuracy of the UCS prediction of rocks, X-ray CT slices pre-processed with the Laplacian of Gaussian (LOG) algorithm were used to enhance the feature sharpness of the compositions in the rocks. Then, a convolutional neural network (CNN) technique based on the stochastic pooling method was applied to the LOG images of the X-ray slices to estimate the UCS of the rocks. This proposed method shows superior performance for the UCS prediction of rocks and could be widely used in the future of artificial intelligence rock engineering.