华丰^1,2,田亮³,邱彤^1,2

• 1:清华大学 化学工程系，北京 100084
• 2:工业大数据系统与应用北京市重点实验室，北京 100084
• 3:中国石油兰州化工研究中心，甘肃 兰州 730060

摘要(Abstract)：

化工过程系统往往具有很强的非线性。针对含有大量序列数据的化工过程建模，当序列数据作为深层神经网络输入时，往往权重系数过多、训练难度增大。而递归神经网络通过在不同时间步间共享参数，更适用于对序列数据的处理。作者研究了递归神经网络在化工动力学建模中的应用，探讨了化学反应中物质浓度的时序变化，反应动力学参数回归，工业油田轻烃裂解过程模拟以及操作条件优化等3种应用场景。从预测精度和计算速度方面，验证了递归神经网络方法在化工过程建模中的优越性。

关键词(KeyWords)： 化学动力学;递归神经网络;过程建模;深度学习

Abstract:

Keywords:

基金项目(Foundation): 国家自然科学基金重点项目（U1462206）。

作者(Author): 华丰^1,2,田亮³,邱彤^1,2

DOI: 10.3969/j.issn.1003-9015.2019.05.021

参考文献(References)：

• [1] VAJDA S. RABIRZ H. Identifiability and distinguishability of general reaction systems[J]. Journal of Physical Chemistry, 1994, 98(20):5265-5271.
  [2] CONNORS K A. Chemical kinetics:The study of reaction rates in solution[M]. New York:John Wiley & Sons, 1990.
  [3] WHITE J E, CATALLO W J, LEGENDRE B L. Biomass pyrolysis kinetics:A comparative critical review with relevant agricultural residue case studies[J]. Journal of Analytical and Applied Pyrolysis, 2001, 91(1):1-33.
  [4] HOUGH B R, BECK D A C, SCHWARTZ D T, et al. Application of machine learning to pyrolysis reaction networks:Reducing model solution time to enable process optimization[J]. Computers & Chemical Engineering, 2017, 104:56-63.
  [5] ZHANG M, LIU X, ZHANG Z. A soft sensor for industrial melt index prediction based on evolutionary extreme learning machine[J]. Computers & Chemical Engineering, 2016, 24:1013-1019.
  [6] SHI B, YANG X, YAN L. Optimization of a crude distillation unit using a combination of wavelet neural network and line-up competition algorithm[J]. Chinese Journal of Chemical Engineering, 2017, 25(8):1013-1021.
  [7] 王玉乔, 程光旭, 汤杰国. 基于KPCA的连续重整催化剂结焦碳含量软测量模型[J]. 高校化学工程学报, 2009, 23(5):852-857. WANG Y Q, CHENG G X, TANG J G. KPCA based soft-sensing model of carbon content of catalyst coking in continuous catalytic reforming reactors[J]. Journal of Chemical Engineering of Chinese Universities, 2009, 23(5):852-857.
  [8] ZHANG J, DAN Q U, ZHEN L I. Recurrent neural network language model based on word vector features[J]. Pattern Recognition & Artificial Intelligence, 2015, 28(4):299-305.
  [9] GUZMAN F, JOTY S, MARQUEZ L, et al. Machine translation evaluation with neural networks[J]. Computer Speech & Language, 2017, 45:180-200.
  [10] OGAWA A, HORI T. Error detection and accuracy estimation in automatic speech recognition using deep bidirectional recurrent neural networks[J]. Speech Communication, 2017, 89:70-83.
  [11] SU H T, MCAVOY T J, WERBOS P. Long-term predictions of chemical processes using recurrent neural networks:A parallel training approach[J]. Industrial & Engineering Chemistry Research, 1992, 31(5):1338-1352.
  [12] 张磊, 陈丙珍, 邱彤. 基于PCA的反应网络简化策略[J]. 化工学报, 2011, 62(1):137-141. ZHANG L, CHEN B Z, QIU T. Reaction network model reduction based on PCA[J]. CIESC Journal, 2011, 62(1):137-141.
  [13] FANG Z, QIU T, CHEN B Z. Improvement of ethylene cracking reaction network with network flow analysis algorithm[J]. Computers & Chemical Engineering, 2016, 91:182-194.
  [14] 胡益锋, 徐用懋, 何细藕. 石脑油裂解炉动态建模与仿真研究[J]. 石油学报(石油加工), 2004, 20(6):51-57. HU Y F, XU Y M, HE X O. The dynamic modeling and simulation of naphtha pyrolysis furnace[J]. Acta Petrolei Sinica (Petroleum Processing Section), 2004, 20(6):51-57.