Permeability Predictions for Tight Sandstone Reservoir Using Explainable Machine Learning and Particle Swarm Optimization

Abstract

High-precision permeability prediction is of great significance to tight sandstone reservoirs. However, while considerable progress has recently been made in the machine learning based prediction of reservoir permeability, the generalization of this approach is limited by weak interpretability. Hence, an interpretable XGBoost model is proposed herein based on particle swarm optimization to predict the permeability of tight sandstone reservoirs with higher accuracy and robust interpretability. The porosity and permeability of 202 core plugs and 6 logging curves (namely, the gamma-ray (GR) curve, the acoustic curve (AC), the spontaneous potential (SP) curve, the caliper (CAL) curve, the deep lateral resistivity (RILD) curve, and eight lateral resistivity (RFOC) curve) are extracted along with three derived variables (i.e., the shale content, the AC slope, and the GR slope) as data sets. Based on the data preprocessing, global and local interpretations are performed according to the Shapley additive explanations (SHAP) analysis, and the redundant features in the data set are screened to identify the porosity, AC, CAL, and GR slope as the four most important features. The particle swarm optimization algorithm is then used to optimize the hyperparameters of the XGBoost model. The prediction results of the PSO-XGBoost model indicate a superior performance compared with that of the benchmark XGBoost model. In addition, the reliable application of the interpretable PSO-XGBoost model in the prediction of tight sandstone reservoir permeability is examined by comparing the results with those of two traditional mathematical regression models, five machine learning models, and three deep learning models. Thus, the interpretable PSO-XGBoost model is shown to have more advantages in permeability prediction along with the lowest root mean square error, thereby confirming the effectiveness and practicability of this method.

Optimizing acidizing design and effectiveness assessment with machine learning for predicting post-acidizing permeability

“…A proposed XGBoost model, optimized with particle swarm optimization, outperforms benchmark models and traditional methods for predicting tight sandstone reservoir permeability, showcasing superior performance. These findings highlight the potential of machine learning for improved permeability prediction in geoscience applications 29 , 30 . Mathematical models for sandstone acidizing were developed in the 1970s, but predicting the outcome of the process remains difficult due to the complexity of porous media and reactions.…”

Novel XAI techniques for explaining GA applications in job scheduling

“…Step 4. Solve the scheduling problem to generate the optimal scheduling plan: AI technologies can be applied to find optimal solutions to mathematical programming models, prepare inputs for dispatching rules, or optimize dispatching rules [19][20][21][22].…”

“…To solve these problems, this study proposes several novel XAI techniques, including decision tree-based interpretation, dynamic transformation and contribution diagrams, and improved bar charts, to improve the effectiveness of explaining the application of GA in job scheduling. The proposed method can also be used to explain the application of other evolutionary AI techniques, such as Ant Colony Optimization (ACO) [20], Particle Swarm Optimization (PSO) [21], Artificial Bee Colony (ABC) [22], job scheduling.…”

Prediction of permeability of highly heterogeneous hydrocarbon reservoir from conventional petrophysical logs using optimized data-driven algorithms

“…When it comes to HMLMs, metaheuristic optimization algorithms have been used to determine hyperparameters of estimator algorithms in the process of training the models. Examples of such models include hybrid ANN-GA 1 and ANN-PSO 2 (Al Khalifah et al 2020;Farouk et al 2021;Kardani et al 2021;Matinkia et al 2022b;Tian et al 2022), PSO-XGBoost (Gu et al 2021;Liu 2022), PSO-SVM (Akande et al 2017;Mahdaviara et al 2020a;Yin et al 2020;Tian et al 2022), hybrid ELM-PSO and ELM-GA (Mahdaviara et al 2020a;Kardani et al 2021), and hybrid RF-PSO and RF-GA (Wang et al 2020;Tian et al 2022). Table 1 summarizes the studies focusing on the prediction of permeability by applying HML and SML models.…”