Application of active learning technique in recognition of diagenetic facies of Mesozoic buried-hill clastic rock reservoirs in Chengdao-Zhuanghai area
The Mesozoic buried-hill clastic rock reservoirs in the Chengdao-Zhuanghai area have great exploration potential， but the reservoirs are highly heterogeneous， with few wells available for coring， making it challenging to predict high-quality reservoirs. A training dataset where logging...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | zho |
| Published: |
Editorial Office of Petroleum Geology and Recovery Efficiency
2025-07-01
|
| Series: | Youqi dizhi yu caishoulu |
| Subjects: | |
| Online Access: | https://yqcs.publish.founderss.cn/thesisDetails#10.13673/j.pgre.202311038&lang=en |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The Mesozoic buried-hill clastic rock reservoirs in the Chengdao-Zhuanghai area have great exploration potential， but the reservoirs are highly heterogeneous， with few wells available for coring， making it challenging to predict high-quality reservoirs. A training dataset where logging data corresponded to diagenetic facies labels was established through the statistical analysis of core samples， thin sections， and logging data， as well as the petrologic characteristics and diagenetic facies types of the reservoir in this region. Recognition of diagenetic facies of the reservoirs was carried out using active learning and supervised learning techniques， and the development intervals of high-quality reservoirs were predicted. The results indicate that most Mesozoic buried-hill clastic rock reservoirs in the Chengdao-Zhuanghai area are composed of lithic feldspar sandstones， followed by feldspathic lithic sandstone. The main reservoir space is composed of secondary pores， followed by fractures. They can be classified into four diagenetic facies types： strong compaction with moderate-weak dissolution and weak cementation facies （SCF）， strong dissolution with strong-moderate compaction and weak cementation facies （SDF）， strong cementation with weak compaction and weak dissolution facies （SCF）， and transitional facies （TF）. Among them， the diagenetic facies named SDF exhibit the best overall reservoir properties， followed by TF. The machine learning results indicate that the active learning technique exhibits strong inter-well generalization capabilities， high prediction accuracy， and enhanced human-machine interactive learning abilities compared to random forest supervised classification models. The TF with moderate preferred physical properties is the primary diagenetic facies type in the Mesozoic buried-hill clastic rock reservoirs in the Chengdao-Zhuanghai area， followed by SDF. Differential dissolution is the primary controlling factor for Mesozoic high-quality reservoirs in the Chengdao-Zhuanghai area. |
|---|---|
| ISSN: | 1009-9603 |