化工学报 ›› 2022, Vol. 73 ›› Issue (1): 342-351.DOI: 10.11949/0438-1157.20211104

• 过程系统工程 • 上一篇    下一篇

基于TA-ConvBiLSTM的化工过程关键工艺参数预测

袁壮1(),凌逸群2,杨哲1(),李传坤1   

  1. 1.中石化安全工程研究院有限公司,化学品安全控制国家重点实验室,山东 青岛 266071
    2.中国石油化工集团有限公司,北京 100728
  • 收稿日期:2021-08-09 修回日期:2021-10-20 出版日期:2022-01-05 发布日期:2022-01-18
  • 通讯作者: 杨哲
  • 作者简介:袁壮(1991—),男,博士,工程师,yuanz.qday@sinopec.com
  • 基金资助:
    国家自然科学基金青年科学基金项目(21706291);中国石化重大科技项目(321123-1)

Critical parameters prediction based on TA-ConvBiLSTM for chemical process

Zhuang YUAN1(),Yiqun LING2,Zhe YANG1(),Chuankun LI1   

  1. 1.State Key Laboratory of Safety and Control for Chemicals, SINOPEC Research Institute of Safety Engineering Co. , Ltd. , Qingdao 266071, Shandong, China
    2.China Petrochemical Corporation, Beijing 100728, China
  • Received:2021-08-09 Revised:2021-10-20 Online:2022-01-05 Published:2022-01-18
  • Contact: Zhe YANG

摘要:

化工过程中,掌握关键工艺参数的变化趋势对于消除潜在波动、维持工况稳定作用巨大。然而,传统的浅层静态模型很难对非线性和动态性显著的复杂序列数据进行精准预测。针对上述难题,提出一种深度预测模型TA-ConvBiLSTM,将卷积神经网络(convolutional neural networks, CNN)和双向长短时记忆网络(bi-directional long short term memory, BiLSTM)集成到统一框架内,使其不仅能在每个时间步上自动挖掘高维变量间的隐含关联,更能横跨所有时间步自适应提取有用的深层时序特征。此外,引入时间注意力(temporal attention, TA)机制,为反映目标变化规律的重要信息增加权重,避免其因输入序列过长、深层特征太多而被掩盖。所提出方法的有效性在国内某延迟焦化装置炉管温度预测的案例中得到验证。

关键词: 化学过程, 预测, 神经网络, 双向长短时记忆网络, 卷积神经网络, 时间注意力机制

Abstract:

In the chemical process, mastering the change trend of key process parameters has a great effect on eliminating potential fluctuations and maintaining stable working conditions. However, traditional shallow static models are difficult to accurately predict complex sequence data with significant nonlinearity and dynamics. In response to the above problems, a deep prediction model called TA-ConvBiLSTM is proposed by seamlessly integrating convolutional neural networks(CNN) and bi-directional long short term memory(BiLSTM) into a unified framework. In this way, the integrated model can, not only automatically explore the esoteric relevance among high-dimensional variables at each time step, but also adaptively extract useful deep temporal features across all time steps. In addition, the temporal attention(TA) mechanism is further introduced to increase the weight of significant information reflecting the law of target variation, so as to prevent it from being concealed due to the overlong input sequence and over many deep features. The effectiveness of the proposed method is verified in a case of furnace tube temperature prediction in a domestic delayed coking unit.

Key words: chemical process, prediction, neural networks, bi-directional long short term memory, convolutional neural networks, temporal attention mechanism

中图分类号: