Full
SUN Yuxi,born in 1997,is a Ph.D. candidate at China University of Petroleum(Beijing). He is mainly engaged in reservoir geology. E-mail: 2860485642@qq.com.
Outline
In recent years,machine learning has shown significant advantages in reservoir parameter evaluation based on well logs. However,due to the limited size of core samples,the generalization ability of these models is often not guaranteed. In this study,more than 100 sets of high-pressure mercury injection data and seven types of well logs were collected from the Triassic Yanchang Formation in the Jiyuan area of the Ordos Basin. The study systematically compares the performance of TabPFN and seven other common machine learning methods under small data conditions. The SHAP interpretability analysis is also used to understand the model’s decision-making mechanisms. The results show: (1)The average permeability of the Yanchang Formation sandstone reservoir in the study area is 0.39×10-3 μm2,which is typical of low- to ultra-low permeability reservoirs. These reservoirs have small pore throat radii and complex pore structures. The first principal component explains 65.00% of the total variance of the pore structure parameters. Combined with production data and mechanism analysis,it is considered an appropriate target variable for characterizing pore structure. (2)The TabPFN model performs excellently in small-data environments,achieving R2 values of 0.81 and 0.87 in the validation and blind well test sets,respectively. Its high performance without parameter tuning demonstrates its outstanding applicability. (3)According to feature importance ranking,density log,sedimentary facies,acoustic log,geological layering,and deep resistivity are the top five features. Sedimentary facies and geological layering make greater contributions in extreme pore structure intervals,highlighting the critical role of discrete geological variables in improving model performance. This study provides an effective data-driven case for characterizing the pore structure of low-permeability sandstone reservoirs under small data conditions.
169
80
| 科 Family | 属数 Number of genus | 种数 Number of species | 占总种数比例 Percentage of total species (%) | 属 Genus | 种数 Number of species | 占总种数比例 Percentage of total species (%) |
|---|---|---|---|---|---|---|
| 鹅膏菌科Amanitaceae | 2 | 11 | 5.26 | 鹅膏菌属 Amanita | 10 | 4.78 |
| 小菇科 Mycenaceae | 2 | 12 | 5.74 | 丝盖伞属 Inocybe | 5 | 2.39 |
| 多孔菌科 Polyporaceae | 8 | 14 | 6.70 | 蜡蘑属 Laccaria | 5 | 2.39 |
| 红菇科 Russulaceae | 3 | 23 | 11.00 | 小皮伞属 Marasmius | 6 | 2.87 |
| 小菇属 Mycena | 11 | 5.26 | ||||
| 光柄菇属 Pluteus | 5 | 2.39 | ||||
| 红菇属 Russula | 17 | 8.13 | ||||
| 栓菌属 Trametes | 5 | 2.39 |
关闭全屏
No. 86 Xueyuan South Road, Haidian District, Beijing
010-62662699(Operation and Technical Support)
qkjq@cast.org.cn
Copyright © 2025 China Association for Science and Technology. All rights reserved. For all open access content, the relevant licensing terms apply.
Sponsored by the Office of the Leading Group for Cybersecurity and Informatization of CAST, and supported by Science and Technology Review Publishing House
