【摘要】 针对当前提高石油采收率技术的传统人工筛选决策方法与现代数据分析决策方法各自的局限性，运用人工智能与数据分析技术，将领域专家知识和机器学习方法有机融合起来，建立基于孪生数据信息的提高石油采收率(EOR)智能决策系统。通过重构提高石油采收率数据信息并进行降噪提质，揭示不同EOR技术的驱油机理及油藏-流体适用条件；利用机器学习探究不同EOR油藏-流体参数权重，构建领域专家知识本体与机器学习推演的孪生数据信息融合与智能决策推理方法。通过Midway Sunset油藏案例验证了所建的基于孪生数据信息的EOR智能决策模型可靠性，可为老油田提高石油采收率技术快捷、科学、高效决策提供一定借鉴。更多还原

【Abstract】 In response to the limitations of conventional artificial methods and advanced data analysis decision-making methods for screening of enhanced oil recovery(EOR) techniques, an intelligent decision-making system for enhanced oil recovery(EOR) based on twin data information was established by combining expert knowledge with machine learning methods using artificial intelligence and data analysis technology.The oil displacement mechanisms and reservoir-fluid applicability conditions of different EOR technologies are revealed by the reconstruction and noise-reduction and quality-improvement of the EOR data information; The reservoir-fluid parameter weights of different EOR techniques are explored by machine learning, and a twin data information fusion and intelligent decision-making reasoning method between domain expert knowledge ontology and machine learning inference are constructed.The reliability of the EOR technology intelligent decision-making model based on twin data information was verified through the Midway Sunset oil reservoir case, which can provide some reference for the quick, scientific, and efficient decision-making of enhanced oil recovery technologies in old oil fields.更多还原