Toshiba AI Technology Catalog

  • Operation and Control

AI collaborative control for increasing efficiency of industrial controller adjustments

Reduce man-hours for front-line adjustments of industrial controllers by adjustment staff.


  • Develop AI collaborative control methods applying conventional controls such as PID control*1 to reinforcement learning*2, reduce unstable operations seen during learning, and recover control performance.  
  • Reductions in manpower required for adjustments can be expected by automating controller adjustments using AI collaborative control.
  • PID (Proportional-Integral-Differential) control: A control method that has gained broad acceptance; adjust three parameters in keeping with the control target.
  • Reinforcement learning: A learning method for evaluating control performance for control targets, to enable improved control.

Applications



  • Industrial controllers, plant controls, device controls

Benchmarks, strengths, and track record



  • Using AI collaborative control method, unstable operations seen during learning were reduced to about 1/3rd.

Inquiries



Toshiba Infrastructure Systems & Solutions Corporation Infrastructure Systems Research and Development Center

Please include the title “Toshiba AI Technology Catalog: AI collaborative control for increasing efficiency of industrial controller adjustments” or the URL in the inquiry text.
Please note that because this technology is currently the subject of R&D activities, immediate responses to inquiries may not be possible.

References:

  • Yang Qin, Toshiya Takano; “Improving the Efficiency of Reinforcement Learning by Utilizing the Operation of Existing Controller”; 2022 Electronics, Information and Systems Conference, pp.1095 – 1100
  • Yang Qin, Toshiya Takano; “Mini Model Verification of Reinforcement Learning by Utilizing the Operation of Existing Controller”; 2023 Electronics, Information and Systems Conference, pp.491 – 496
  • Toshiya Takano, Yang Qin; ”Optimal Implementation Approach to Embedded System for Reinforcement Learning Utilizing Manipulation of Existing Controller”; 2023 Electronics, Information and Systems Conference, pp.497 – 502
  • “Control technology for reinforcement learning to reduce learning time and improve control performance”; TOSHIBA REVIEW SCIENCE AND TECHNOLOGY HIGHLIGHTS 2024, pp.25
    TOSHIBA REVIEW Science and Technology Highlights 2024 (PDF)(3.03MB)