Data-driven modeling of refinery reactors

炼油反应器的数据驱动建模

• 批准号: 533718-2018
• 负责人: Reformat, Marek
• 金额: $ 1.82万
• 依托单位: University of Alberta
• 依托单位国家: 加拿大
• 项目类别: Engage Grants Program
• 财政年份: 2018
• 资助国家: 加拿大
• 起止时间: 2018-01-01 至 2019-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=663124
• 关键词: Data; driven; modeling; refinery; reactors

A rapid development and increased popularity of data processing methods - from statistical analysis via machine leaning to computational intelligence and deep learning - have triggered an interest in utilizing new approaches to modeling processes. It is expected that applications of techniques of artificial intelligence in development of new generation of data driven models will lead to an enhanced utilization of models in design and validation of more efficient and optimal industrial control systems. Advanced data-oriented modelling methods will be able to address many issues related to systems' performance and their improvements.**For these reasons, the project focuses on development of methods and techniques for better understanding of factors influencing modeling of physical process units in refineries. It is important to utilize fast and accurate models of refineries in cases where currently used models require high performance computing, and are practically useless for quick predictability activities and processes of finding the best working configurations for given circumstances. Such knowledge is necessary to develop strategies and methodologies required by new methods of controlling and optimizing operations of refineries.**

数据处理方法的快速发展和日益普及-从统计分析到机器学习到计算智能和深度学习-引发了人们对利用新方法建模过程的兴趣。预计人工智能技术在开发新一代数据驱动模型中的应用将导致模型在设计和验证更有效和最佳工业控制系统中的利用率提高。先进的面向数据的建模方法将能够解决与系统性能及其改进有关的许多问题。由于这些原因，该项目的重点是开发方法和技术，以更好地了解影响炼油厂物理过程单元建模的因素。在当前使用的模型需要高性能计算的情况下，利用炼油厂的快速和准确的模型是重要的，并且对于快速预测活动和针对给定情况找到最佳工作配置的过程实际上是无用的。这种知识对于制定控制和优化炼油厂操作的新方法所需的战略和方法是必要的。