人工智能技术在气藏工程专业的应用前景展望

Artificial intelligence techniques and their application to gas reservoir engineering

气藏工程研究涉及诸多复杂任务,模拟人类思维进行诊断分析、预测评估、认知推理、决策优化的智能技术研发尚处于起步阶段,亟须解决找准典型应用场景、明确人机协同分工、优选攻关方向的问题。为此,基于长期从事气藏工程理论研究与应用实践形成的认识,以及四川盆地气田开发智能化初步探索的成功经验,开展气藏工程新一代人工智能技术研发的需求分析和可行性评价,由此聚焦重点、展望技术发展前景。研究结果表明:①气藏工程研究对人工智能的需求主要表现在处理复杂事务时提高效率、不确定性较强的条件下保障分析质量以及人机协同挖掘规律性认识等三个方面;②面向上述需求,系统地开展本专业智能化转型的专项攻关才能解决深层次技术问题;③自动分析预测、智能聚类判识、大数据驱动知识学习、人机协同增强智能是气藏工程专业人工智能的研究重点。结论认为,新型人机协同是气藏工程研究的发展趋势。通过梳理气藏工程专业对人工智能技术的需求和相关问题,初步明确了攻关方向,能为后续研究提供参考。

There are numerous complex tasks to study gas reservoir engineering.Both research and development(R&D)of artificial intelligence(AI)techniques,that simulate human thinking for diagnosis and analysis,prediction and evaluation,cognition and reasoning,and decision making and optimization,are still in the stage of beginning.And it is urgent to solve some problems of capturing typical application scenarios,definiting human-computer collaboration and division,and optimizing a few tackling directions.Thus,based on theortical researches on gas reservoir engineering and understandings achieved from application practices for a long time,as well as successful probing experiences on intelligent gasfield construction in Sichuan Basin,R&D demands were analyzed and their feasibility was evaluated for new generation of AI techniques used in gas reservoir engineering.In addition,the key points and prospects for technological application were proposed.Results show that(1)gas reservoir engineering studies mainly need AI techniques to improve their efficiency when dealing with complexities,ensure analysis quality with emerged stronger uncertainty,and figure out regularities through human-computer collaboration;(2)for the above requirements,only to systematically implement unique researches on discipline-oriented intelligent transformation can solve deep-seated technical problems;and(3)automatic analysis and prediction,smart clustering and identification,big data-driven knowledge learning and human-computer collaboration to boost intelligence are cores in AI-related researches on gas reservoir engineering.It is concluded that novel human-computer collaboration is the next trend to study gas reservoir engineering.In this paper,both needs and related issues about AI techniques are presented for gas reservoir engineering and the research directions are initially clarified,which may provide reference for future efforts.