化工进展 ›› 2021, Vol. 40 ›› Issue (4): 1765-1776.DOI: 10.16085/j.issn.1000-6613.2020-2139

• 专栏:先进化工装备及智能系统工程 • 上一篇    下一篇

化工过程的智能混合建模方法及应用

张梦轩(), 刘洪辰, 王敏, 蓝兴英(), 石孝刚, 高金森   

  1. 中国石油大学(北京)重质油国家重点实验室,北京 102249
  • 收稿日期:2020-10-26 出版日期:2021-04-05 发布日期:2021-04-14
  • 通讯作者: 蓝兴英
  • 作者简介:张梦轩(1992—),男,博士研究生。E-mail:christzmx@163.com
  • 基金资助:
    国家自然科学基金(91834303);中国石油大学(北京)科研基金(2462018BJC003)

Intelligence hybrid modeling method and applications in chemical process

ZHANG Mengxuan(), LIU Hongchen, WANG Min, LAN Xingying(), SHI Xiaogang, GAO Jinsen   

  1. State Key Laboratory of Heavy Oil Processing, China University of Petroleum, Beijing 1002249, China
  • Received:2020-10-26 Online:2021-04-05 Published:2021-04-14
  • Contact: LAN Xingying

摘要:

随着人工智能技术和配套数据系统的快速发展,化工过程建模技术达到了新的高度,将多个机理模型和数据驱动模型以合理的结构加以组合的智能混合建模方法,可以综合利用化工过程的第一性原理及过程数据,结合人工智能算法以串联、并联或者混联的形式解决化工过程中的模拟、监测、优化和预测等问题,建模目的明确,过程灵活,形成的混合模型有着更好的整体性能,是近年来过程建模技术的重要发展趋势。本文围绕近年来针对化工过程的智能混合建模工作进行了总结,包括应用的机器学习算法、混合结构设计、结构选择等关键问题,重点论述了混合模型在不同任务场景下的应用。指出混合建模的关键在于问题和模型结构的匹配,而提高机理子模型性能,获取高质量宽范围的数据,深化对过程机理的理解,形成更有效率的混合建模范式,这些都是现阶段提高混合建模性能的研究方向。

关键词: 过程系统工程, 混合建模, 机器学习, 人工智能, 神经网络

Abstract:

With the rapid development of artificial intelligence and the corresponding supporting data system, chemical process modeling technology has been developed to a new and high level. The hybrid modeling method combining multiple mechanism models and data-driven models in a reasonable structure can fully use the first principle and process data of chemical process. With the help of artificial intelligence algorithms, the hybrid modeling can achieve simulation, monitoring, optimization, and prediction in chemical process in the form of series, parallel or hybrid. It is an important development trend for process modeling technology in recent years because the modeling purpose is clear, the model development is flexible, and the resulting hybrid model is in high performance. This work summarized the recent advances on intelligent hybrid modeling in chemical process, including the application of machine learning algorithms, hybrid structure design, and structure selection. The application of hybrid model was discussed in different task scenarios. Through literature review and analysis, it was concluded that the key of hybrid modeling lies in the matching of problem and model structure. More fruitful results can be achieved for the hybrid modeling technology by improving the performance of mechanism sub-models, acquiring high quality data in a wide range, deepening the understanding of process mechanism, and establishing a more efficient hybrid modeling paradigm.

Key words: process systems engineering, hybrid modeling, machine learning, artificial intelligence, neural network

中图分类号: 

京ICP备12046843号-2;京公网安备 11010102001994号
版权所有 © 《化工进展》编辑部
地址:北京市东城区青年湖南街13号 邮编:100011
电子信箱:hgjz@cip.com.cn
本系统由北京玛格泰克科技发展有限公司设计开发 技术支持:support@magtech.com.cn