AI and Simulation for Soft Sensors and Process Control

DOI Web Site Web Site 10 References
  • Kubosawa Shumpei
    NEC-AIST AI Cooperative Research Laboratory, National Institute of Advanced Industrial Science and Technology Data Science Research Laboratories, NEC Corporation
  • Onishi Takashi
    NEC-AIST AI Cooperative Research Laboratory, National Institute of Advanced Industrial Science and Technology Data Science Research Laboratories, NEC Corporation
  • Tsuruoka Yoshimasa
    NEC-AIST AI Cooperative Research Laboratory, National Institute of Advanced Industrial Science and Technology Department of Information and Communication Engineering, Graduate School of Information Science and Technology, The University of Tokyo

Bibliographic Information

Other Title
  • AIとシミュレーションによるソフトセンサとプロセス制御
  • AI ト シミュレーション ニ ヨル ソフトセンサ ト プロセス セイギョ

Search this article

Abstract

<p>In the operation of a chemical plant, the quality of the products must be maintained at a constant level and the production of off-specification products should be minimized. For this purpose, it is necessary to measure process variables related to product quality, such as temperature and composition of materials at various parts of the plant, and perform appropriate operations (i.e. control) based on the measurements. Some process variables, such as temperature and flow rate, can be measured continuously and instantaneously; however, other variables, such as composition and viscosity, can only be obtained through time-consuming analysis after sampling substances from the plant. Soft sensors are proposed to estimate process variables that cannot be obtained in real-time from easy-to-measure variables. However, in the statistically constructed soft sensors from recorded measurements, the estimation accuracy in unrecorded situations (extrapolation) is not guaranteed and deteriorated in general. In this paper, to improve the extrapolation performance, we propose to use internal state variables of a plant as soft sensors by estimating them using a dynamic simulator that can estimate and predict even unrecorded situations based on chemical engineering knowledge, and reinforcement learning, one of the AI methods. Additionally, we describe the prospects for plant operation and control using such soft sensors and the methodology to obtain the prediction models (i.e. simulators) required for the proposed system.</p>

Journal

  • KAGAKU KOGAKU RONBUNSHU

    KAGAKU KOGAKU RONBUNSHU 48 (4), 141-151, 2022-07-20

    The Society of Chemical Engineers, Japan

References(10)*help

See more

Keywords

Details 詳細情報について

Report a problem

Back to top