中国石油勘探 ›› 2020, Vol. 25 ›› Issue (5): 89-96.DOI: 10.3969/j.issn.1672-7703.2020.05.012

• 工程技术 • 上一篇    下一篇

基于梦想云的人工智能地震解释模式研究与实践

杨 平,詹仕凡,李 明,李 磊,郭 锐,尚民强,陶春峰   

  1. 中国石油集团东方地球物理勘探有限责任公司
  • 出版日期:2020-09-14 发布日期:2020-09-12
  • 基金资助:
    中国石油天然气集团有限公司科学研究与技术开发项目“深层与非常规物探新方法新技术”(2019A-33);中国石油天 然气股份有限公司投资信息化重点项目“勘探开发一体化协同研究及应用平台(一期)建设”(PetroChina-IT-2017-N104);中国石 油集团东方地球物理勘探有限责任公司科技项目“储层高分辨地震反演技术深化研究及软件集成”(01-03-02-2019)。

Research and practice on an artificial intelligence seismic interpretation mode based on the E&P Dream Cloud

Yang Ping, Zhan Shifan, Li Ming, Li Lei, Guo Rui, Shang Minqiang, Tao Chunfeng   

  1. BGP Inc., China National Petroleum Corporation
  • Online:2020-09-14 Published:2020-09-12
  • Supported by:
     

摘要: 为了探索云计算与人工智能技术在地震资料解释领域的结合方式,更好地发挥两者的协同优势,在深度 学习解释软件研发实践的基础上,提出了基于梦想云+深度学习方法的地震资料解释新模式:①由梦想云数据湖与 计算中心为深度学习算法提供大量的高质量标签数据与高弹性计算平台;②深度学习技术与传统地球物理方法优势互 补,共同搭建地震与地质解释的新流程;③在梦想云平台上应用深度学习模块,直接为梦想云平台的业务决策场景提 供有效地震成果信息支持。这种模式可以取代大量手工操作、实现更高预测精度,从而为高效益、高精度油气勘探打 下坚实的基础,经实践验证,取得了令人鼓舞的效果。

 

关键词: 梦想云, 深度学习, 人工智能, 地震资料解释

Abstract: A new seismic data interpretation mode has been proposed, which is based on the E&P Dream Cloud and deep learning methods. This new approach explores the combination of cloud computing and artificial intelligence technology in seismic data interpretation, taking advantage of their collaborative capabilities. It is based on practical R&D experience with deep-learning seismic interpretation software, which includes: (1) Data Lake and Computing Center of the E&P Dream Cloud provides a large volume of high-quality labeled data and a highly elastic computing platform on which to run the deep learning algorithm. (2) Deep-learning technology and traditional geophysical methods complement each other to establish a new process for seismic and geological interpretation. (3) The deep-learning module in the E&P Dream Cloud platform provides effective information support from seismic results for business decision-making within the E&P Dream Cloud. This mode considerably reduces the number of manual operations required and achieves higher prediction accuracy, laying a solid foundation for high-efficiency and high-precision oil and gas exploration. Experience in practice with real data has already produced encouraging results.

Key words: E&, P Dream Cloud, deep learning, artificial intelligence, seismic data interpretation

中图分类号: