ABSTRACT
The prediction of geological and geophysical parameters for well productivity is essential in optimizing exploration, resource distribution, and sustainable energy extraction. However, the inherent complexities of geoscience datasets, characterized by high-dimensionality, missing values, and non-linear feature interactions, pose significant drawbacks to conventional modeling approaches. This study proposes GeoStack-AI, a novel hybrid ensemble framework that incorporates advanced machine learning models such as Random Forest Gradient Boosting and XGBoost within a stacking ensemble framework to improve predictive performance and interpretability. The framework incorporates advanced feature engineering approach to capture complex interactions and achieve dimensionality reduction. Explainability AI technique, specifically SHapley Additive exPlanations (SHAP) are leveraged to offer transparency into feature contributions and support informed insights for decision-making. GeoStack-AI is validated using geological dataset containing well log features and production metrics. Results show that the optimized stacking ensemble model achieved a satisfactory enhancement in predictive performance, with a test R2 score of 90.9%, outweighing individual base models and conventional ensemble approaches. Top influencing features such as Latitude, Township, and Longitude age analyzed SHAP, providing insights into the essential role of spatial feature in well productivity. The framework also highlights intricate feature interactions between Latitude and Range Direction, suggesting deeper intuitions into geophysical dependencies. GeoStack-AI demonstrates state-of-the-art performance by providing interpretable approach for optimizing resource extraction. The adaptability of GeoStack-AI makes it applicable to other domains such as renewable energy and agriculture.
References
- [1] Abdeljaber, O. Avci, S. Kiranyaz, M. Gabbouj, and D. Inman, “Real-time vibration-based structural damage detection using one-dimensional convolutional neural networks,” J. Sound Vib., vol. 388, pp. 154–170, 2017.
- [2] Abdulaziz, A. Hameeda, and H. Mohamed, “Prediction of reservoir quality using well logs and seismic attributes analysis with artificial neural network: A case study from Farrud reservoir, Al-Ghani field, Libya,” J. Appl. Geophys., vol. 161, pp. 239–254, 2019.
- [3] Ai-Anazi and I. Gates, “Support vector regression to predict porosity and permeability: effect of sample size,” Comput. Geosci., vol. 39, no. 2, pp. 64–76, 2011.
- [4] Alhussein and G. Muhammad, “Voice pathology detection using deep learning on mobile healthcare framework,” IEEE Access, vol. 6, no. 99, pp. 41034–41041, 2018.
- [5] Ali and S. Ebrahim, “Physical properties modeling of reservoirs in Mansuri oil field, Zagros region, Iran,” Petrol. Explor. Dev., vol. 43, no. 4, pp. 559–563, 2016.
- [6] An, D. Cao, B. Zhao, X. Yang, and M. Zhang, “Reservoir physical parameters prediction based on LSTM recurrent neural network,” Prog. Geophys., vol. 34, no. 5, pp. 1849–1858, 2019.
- [7] Azamifard, M. Ahmadi, F. Rashidi, M. Pourfard, and B. Dabir, “MPS realization selection with an innovative LSTM tool,” J. Appl. Geophys., vol. 179, art. no. 104107, 2020.
- [8] Bai, J. Z. Kolter, and V. Koltun, “An empirical evaluation of generic convolutional and recurrent networks for sequence modeling,” arXiv preprint, arXiv:1803.01271, 2018.
- [9] J. Bergen, P. A. Johnson, M. de Hoop, and G. Beroza, “Machine learning for data-driven discovery in solid Earth geoscience,” Science, vol. 363, no. 6433, art. no. eaau0323, 2019.
- Camila, G. Leonardo, and O. Costa, “Machine learning approaches for petrographic classification of carbonate-siliciclastic rocks using well logs and textural information,” J. Appl. Geophys., vol. 155, pp. 217–225, 2018.
- Chen and D. Zhang, “Physics-constrained deep learning of geomechanical logs,” IEEE Trans. Geosci. Remote Sensing, vol. 58, no. 8, pp. 5932–5943, 2020.
- Chen, L. Yang, B. Zha, M. Zhang, and Y. Chen, “Deep learning reservoir porosity prediction based on multilayer long short-term memory network,” Geophysics, vol. 85, no. 4, pp. WA213–WA225, 2020.
- Di, Z. Li, H. Maniar, and A. Abubakar, “Seismic stratigraphy interpretation by deep convolutional neural networks: A semisupervised workflow,” Geophysics, vol. 85, no. 4, pp. WA77–WA86, 2020.
- Fabien-Ouellet and R. Sarkar, “Seismic velocity estimation: A deep recurrent neural-network approach,” Geophysics, vol. 85, no. 1, pp. U21–U29, 2020.
- Feng, “Lithofacies classification based on a hybrid system of artificial neural networks and hidden Markov models,” Geophys. J. Int., vol. 221, no. 3, pp. 1484–1498, 2020.
- Feng, “Estimation of reservoir porosity based on seismic inversion results using deep learning methods,” J. Nat. Gas Sci. Eng., vol. 77, art. no. 103270, 2020.
- Gu, Z. Bao, and G. Cui, “Permeability prediction using hybrid techniques of continuous restricted Boltzmann machine, particle swarm optimization and support vector regression,” J. Nat. Gas Sci. Eng., vol. 59, pp. 97–115, 2018.
- S. U. Islam, “Using deep learning-based methods to classify salt bodies in seismic images,” J. Appl. Geophys., vol. 178, art. no. 104054, 2020.
- Iturrar and J. Parra, “Artificial neural networks applied to estimate permeability, porosity and intrinsic attenuation using seismic attributes and well-log data,” J. Appl. Geophys., vol. 107, no. 8, pp. 45–54, 2014.
- Khansari and C. Griffiths, “Porosity from sonic log in gas-bearing shale sandstones: field data versus empirical equations,” Explor. Geophys., vol. 29, no. 4, pp. 440–446, 1998.
Download all article in PDF
