Physics-Guided Machine Learning for Transient Reservoir Characterization: Synthetic Dataset, Models, Validation and Benchmark Results

This dataset supports the study “Physics-Guided Machine Learning for Transient Reservoir Characterization: Accuracy, Out-of-Distribution Generalization, and Noise Robustness in Niger Delta Sandstones.” It contains synthetic pressure-transient reservoir data generated across representative sandstone reservoir and fluid-property ranges, together with variables used for physics-guided machine-learning model development, training, validation, testing, out-of-distribution generalization assessment, and noise-robustness evaluation. The dataset includes reservoir and fluid parameters, transient pressure-response features, and corresponding target reservoir properties. It is provided to support reproducibility, independent verification, and further research on machine-learning-assisted transient reservoir characterization.本数据集支撑研究《物理引导机器学习用于油藏瞬态表征：尼日尔三角洲砂岩中的精度、分布外泛化与噪声鲁棒性》（Physics-Guided Machine Learning for Transient Reservoir Characterization: Accuracy, Out-of-Distribution Generalization, and Noise Robustness in Niger Delta Sandstones）。其包含基于典型砂岩油藏与流体物性范围生成的合成油藏瞬态压力数据，以及用于物理引导机器学习（Physics-Guided Machine Learning）模型开发、训练、验证、测试、分布外泛化评估与噪声鲁棒性评估的相关变量。本数据集涵盖油藏与流体参数、瞬态压力响应特征，以及对应的目标油藏属性。本数据集的发布旨在支持可复现性研究、独立验证，以及机器学习辅助油藏瞬态表征领域的后续研究工作。