Page 79 - Machine Learning for Subsurface Characterization

Chapter 3

Shallow neural networks and

classification methods for
approximating the subsurface

in situ fluid-filled pore size
distribution

Siddharth Misra* and Jiabo He †,a
*
Harold Vance Department of Petroleum Engineering, Texas A&M University, College Station,
†
TX, United States, School of Computing and Information Systems, University of Melbourne,
Parkville, VIC, Australia
Chapter outline
1 Introduction 66 3 ANN model training, testing, and
2 Methodology 67 deployment 81
2.1 Hydrocarbon-bearing shale 3.1 ANN models 81
system 67 3.2 Training the first ANN model 81
2.2 Petrophysical basis for the 3.3 Testing the first ANN model 84
proposed data-driven log 3.4 Training the second ANN
synthesis 68 model 84
2.3 Data preparation and 3.5 Testing the second ANN
statistical information 69 model 86
2.4 Categorization of depths 3.6 Petrophysical validation
using flags 72 of the first ANN model 86
2.5 Fitting the T 2 distribution 3.7 ANN-based predictions
with a bimodal Gaussian of NMR T 2 distribution for
distribution 74 various depth intervals 87
2.6 Min-max scaling of the 4 Conclusions 89
dataset (features and target) 77 Appendix A Statistical properties
2.7 Training and testing of conventional logs and
methodology for the ANN inversion-derived logs for various
models 78 depth intervals 89

a
. Formerly at the University of Oklahoma, Norman, OK, United States
Machine Learning for Subsurface Characterization. https://doi.org/10.1016/B978-0-12-817736-5.00003-X
© 2020 Elsevier Inc. All rights reserved. 65