• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

Research on Operator Support Functions for Process Industries Using Deep Learning Technology

Research Project

Project/Area Number 19K04113
Research Category

Grant-in-Aid for Scientific Research (C)

Allocation TypeMulti-year Fund
Section一般
Review Section Basic Section 18020:Manufacturing and production engineering-related
Research InstitutionWaseda University

Principal Investigator

Fujimura Shigeru  早稲田大学, 理工学術院(情報生産システム研究科・センター), 教授 (00367179)

Project Period (FY) 2019-04-01 – 2022-03-31
Project Status Completed (Fiscal Year 2021)
Budget Amount *help
¥4,290,000 (Direct Cost: ¥3,300,000、Indirect Cost: ¥990,000)
Fiscal Year 2021: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
Fiscal Year 2020: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Fiscal Year 2019: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
Keywords異常診断 / 時系列予測 / プロセス監視 / 深層学習 / 機械学習 / ディープラーニング / オペレータ支援 / 時系列データ予測 / プロセス産業 / 時系列データ
Outline of Research at the Start

第4次産業革命の実現やIoT(Internet of Things)の導入を日本のプロセス産業で成功させるため,プロセス制御監視システムによって蓄積された時々刻々変化する時系列データを利用しディープラーニング技術を応用した実プロセスで利用可能なオペレータ支援機能を実現する.本研究課題のアプローチは,正常時の複数の時系列データを入力情報とし,1時系列データの将来の挙動を予測するものである.現場力を重んじる日本のプロセス産業において,熟練オペレータに対するポカミス防止,新人オペレータに対するプロセス知識において気づきを与える予測表示機能によるオペレータ支援機能を実現する.

Outline of Final Research Achievements

In this research, we realized an operator support function that can be used in actual processes by applying deep learning technology to time-series data accumulated by a process control monitoring system. Specifically, we proposed a new deep learning model that predicts multiple sensor data of an actual chemical process. The model learns complex relationships among the sensor data being monitored for chemical process control. The model utilizes various time-length influence relationships between related sensor data to implement normal value prediction for a single sensor data.
The developed model has realized an operator support function that prevents errors by experienced operators and provides awareness of process knowledge to new operators. We proposed a method to construct a model using deep learning based on a huge amount of normal process time-series data, and realized a customization-less system construction method that automatically constructs the system.

Academic Significance and Societal Importance of the Research Achievements

正常時の複数の時系列データを入力情報とし、時系列データの将来の挙動を予測するモデルを開発し、熟練オペレータに対するポカミス防止、新人オペレータに対するプロセス知識における気づきを与えるオペレータ支援機能を実現した。膨大な正常時のプロセス時系列データを利用してディープラーニングによってモデルを構築する方法を提案し自動的にシステムを構築するカスタマイズレスなシステム構築手法を実現した。

Report

(4 results)
  • 2021 Annual Research Report   Final Research Report ( PDF )
  • 2020 Research-status Report
  • 2019 Research-status Report
  • Research Products

    (2 results)

All 2021

All Journal Article (1 results) (of which Peer Reviewed: 1 results) Presentation (1 results) (of which Int'l Joint Research: 1 results)

  • [Journal Article] Capturing combination patterns of long- and short-term dependencies in multivariate time series forecasting2021

    • Author(s)
      Song Wen、Fujimura Shigeru
    • Journal Title

      Neurocomputing

      Volume: 464 Pages: 72-82

    • DOI

      10.1016/j.neucom.2021.08.100

    • Related Report
      2021 Annual Research Report
    • Peer Reviewed
  • [Presentation] Sensor Data Prediction in Process Industry by Capturing Mixed Length of Time Dependencies2021

    • Author(s)
      Wen Song, Shigeru Fujimura
    • Organizer
      2021 IEE International Conference on Industrial Engineering and Engineering Management (IEEM)
    • Related Report
      2021 Annual Research Report
    • Int'l Joint Research

URL: 

Published: 2019-04-18   Modified: 2023-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi