无线电通信技术2025，Vol.51Issue(3)：447-453,7.DOI:10.3969/j.issn.1003-3114.2025.03.003
知识迁移驱动的跨油田存储分类模型
Knowledge Transfer-driven Cross-oilfield Storage Classification Model
摘要
Abstract
Knowledge in oil and gas data is essential for automated reservoir classification,and properly managing and utilizing oil and gas data can help oil companies make the best decisions efficiently and significantly reduce costs.Existing methods mainly focus on reservoir classification in a single geological block,but they are not effective in new blocks.It is a very important and challenging prob-lem of how to transfer subsurface characteristics across geological fields and classify reservoirs accurately.A Knowledge Transfer-driven Cross-oilfield Storage(KTCS)classification model,is proposed,which includes a Multi-scale Sensor Extraction(MSE)module for ex-tracting multi-scale feature representations of geological features from multi-variable logging curves.In addition,a Specific Feature Learning(SFL)module is designed to utilize specific information from different fields.Then,a Knowledge Attention Transfer(KAT)module is designed to learn invariant feature representations and transfer geological knowledge from source field to target field.The pro-posed KTCS model is evaluated through extensive experiments on real industrial datasets,when compared with the non-migration meth-ods,F1 score of the proposed KTCS model is at least 15.7%higher.关键词
储层分类/知识迁移/多尺度传感器/地质知识/工业数据集Key words
reservoir classification/knowledge transfer/multi-scale sensor/geological knowledge/industrial datasets分类
信息技术与安全科学引用本文复制引用
贡玉军..知识迁移驱动的跨油田存储分类模型[J].无线电通信技术,2025,51(3):447-453,7.基金项目
国家自然科学基金(62272066) National Natural Science Foundation of China(62272066) （62272066）
