Developing Soft-Sensor Models Using Latent Dynamic Variational Autoencoders

被引:3
作者
Lee, Yi Shan [1 ]
Ooi, Sai Kit [1 ]
Tanny, Dave [1 ]
Chen, Junghui [1 ]
机构
[1] Chung Yuan Christian Univ, Dept Chem Engn, Taoyuan, Taiwan
来源
IFAC PAPERSONLINE | 2021年 / 54卷 / 03期
关键词
Dynamic Nonlinear Process; Soft-Sensor Prediction; Supervised; Variational Autoencoder;
D O I
10.1016/j.ifacol.2021.08.219
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Quality variables, which are usually measured offline, play important roles in describing process behaviors. However, online data obtained from soft sensors are significant as they provide accurate and immediate information. The reliability of online soft sensors is questionable due to changes in sensors, equipment, raw material availability, and operation conditions. In addition, chemical plants have dynamic properties and complex correlations amidst a large number of process variables. This causes most of the predictions obtained from steady-state soft sensors to be inaccurate in representing the particular chemical process. In this paper, the latent dynamic variational autoencoder is proposed to provide an estimation model and supervise soft-sensors. The input data are encoded in the latent space to remove underlying noises and disturbances in the data. Afterward, the dynamical properties are learned in the latent space through the bi-directional recurrent neural network, whose output (latent variable) is used to reconstruct back the input data. A simulation case study is conducted to show the effectiveness of the proposed method. Copyright (C) 2021 The Authors.
引用
收藏
页码:61 / 66
页数:6
相关论文
共 11 条
  • [1] Adaptive recurrent neural network with Lyapunov stability learning rules for robot dynamic terms identification
    Agand, Pedram
    Shoorehdeli, Mahdi Aliyari
    Khaki-Sedigh, Ali
    [J]. ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE, 2017, 65 : 1 - 11
  • [2] [Anonymous], 2015, Neural Information Processing Systems (NIPS)
  • [3] [Anonymous], 2017, A Disentangled Recognition and Nonlinear Dynamics Model for Unsupervised Learning, in Advances in Neural Information Processing Systems
  • [4] Chen R.-Q., 2019, JOINT MODEL ANOMALY
  • [5] Habibie Ikhsanul, 2017, BMVC
  • [6] On-line batch process monitoring using batch dynamic kernel principal component analysis
    Jia, Mingxing
    Chu, Fei
    Wang, Fuli
    Wang, Wei
    [J]. CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS, 2010, 101 (02) : 110 - 122
  • [7] Quality-related fault detection approach based on dynamic kernel partial least squares
    Jia, Qilong
    Zhang, Yingwei
    [J]. CHEMICAL ENGINEERING RESEARCH & DESIGN, 2016, 106 : 242 - 252
  • [8] Inferential control of distillation composition using partial least squares regression
    Kano, M
    Miyazaki, K
    Hasebe, S
    Hashimoto, I
    [J]. KAGAKU KOGAKU RONBUNSHU, 1998, 24 (03) : 425 - 430
  • [9] Karl M., 2017, INT C LEARN REPR ICL
  • [10] Krishnan R. G., 2015, DEEP KALMAN FILTERS, V2000, P1