Adaptive Residual-Gated Two-Stage Ensemble for Porosity Prediction Using Multi-Property Petrophysical Parameters
Abstract
sparse. In formations with mixed lithologies and nonlinear responses of petrophysical properties, empirical formulas and simple regressions
may deviate substantially. Based on the provided engineering codebase, this paper develops an Adaptive Residual-Gated Two-Stage ensemble
(ARG-TS) for porosity prediction. The framework integrates lightweight preprocessing, lithology-aware modeling, and two-stage learning: a
robust GBDT baseline with K-fold out-of-fold calibration, followed by SVR-based residual learning with an error-amplitude-driven sigmoid
gate for adaptive correction. On a dataset with 300 training samples and 43 test samples, the fused model improves test performance from
R=0.947 to 0.963, reducing RMSE from 0.255 to 0.212 and MAE from 0.163 to 0.153. The results show that ARG-TS provides an effective
and extensible solution for small-sample porosity regression using multi-property petrophysical parameters.
Keywords
Full Text:
PDFReferences
[1] Li Ji 'an.Application of Artificial Intelligence Neural Network in Lithology Identification, Porosity and Permeability Prediction:A Case
Study of the Shihongtan Uranium Deposit [J].Northwest Geology, 2010, 43(2): 32-37.
[2] Friedman J H. Greedy function approximation:A gradient boosting machine[J]. The Annals of Statistics, 2001, 29(5): 1189-1232.
[3] Chen T, Guestrin C. XGBoost:A scalable tree boosting system[C]//Proceedings of the 22nd ACM SIGKDD International Conference on
Knowledge Discovery and Data Mining (KDD'16). 2016: 785-794.
[4] Smola A J, SchlkopfB.A tutorial on support vector regression[J]. Statistics and Computing, 2004, 14: 199-222.
[5] Huber P J. Robust estimation of a location parameter[J]. The Annals of Mathematical Statistics, 1964, 35(1): 73-101.
[6] Wang Jun, Cao Junxing,You Jiachun, et al. Prediction of reservoir pore saturation Parameters Based on gated recurrent Unit Neural Net
work [J]. Geophysical Prospecting for Petroleum, 2020, 59(4): 616-627.
[7] Wang Jun, Cao Junxing, Zhou Xin Reservoir porosity prediction Based on Deep Bidirectional recurrent Neural Network [J]. Progress in
Geophysics, 2022, 37(1): 267-274.
[8] Yang Liuqing, Chen Wei, Zha Bei. Research and Application of Prediction of Reservoir Porosity Using Convolutional Neural Network [J].
Progress in Geophysics, 2019, 34(4): 1548-1555.
[9] Vapnik V N. The Nature of Statistical Learning Theory[M].New York: Springer, 1995.
[10] Hastie T, Tibshirani R, Friedman J. The Elements of Statistical Learning: Data Mining, Inference, and Prediction[M]. 2nd ed.New York:
Springer, 2009.
DOI: http://dx.doi.org/10.70711/itr.v3i3.9235
Refbacks
- There are currently no refbacks.
