Select Language : English Japanese
Global Site : NTT Ltd.
Select Language : English 日本語


Brazil: English / 日本語 / Portugues do Brasil
United States: English

Europe, Middle East, Africa (EMEA)

(Belgium, France, Germany, Netherlands, Spain, United Arab Emirates, United Kingdom)
English / 日本語 / Espanol / Deutsch / Francais
Russia: English / 日本語 / русский

Asia Pacific

Japan(Business): English / 日本語
Japan(Residential): English / 日本語
Australia(NTT Com ICT Solutions): English
Mainland China: English / 日本語 / 簡體中文
Hong Kong & Macao: English / 日本語 / 繁体中文 / 簡體中文
India: English / 日本語
Indonesia: English
Korea: English / 日本語 / ???
Malaysia: English
Philippines(DTSI): English
Singapore: English / 日本語
Taiwan: English / 日本語 / 繁体中文
Thailand: English / 日本語
Vietnam: English / 日本語

CASE of Re-connect X 01

AI Solution to Assist the Operations of Complex Chemical Plants

AI Solution to Assist the Operations of
Complex Chemical Plants

Handing Down Skills in Chemical Plants

The chemical industry in Japan is shifting from single-product,mass production to high-mix, low-volume production and variable-mix, variable-volume production in response to increasing globalization and environmental concerns. As a result, chemical plants require advanced controls beyond the capabilities of existing automatic control technology and in some cases require manual operation.
Japan is shifting to the production of highly sophisticated products while depending on more advanced production technology in the face of fierce international competition. However, difficulties lie ahead in securing operators due to the falling birthrate and aging population, and the transfer of skills in chemical plants has become a critical issue for maintaining the strengths of Japan’s chemical industry.


Digital Twinning of Skilled Operator Behavior

Since 2007, NTT Communications has been collaborating with Yokogawa Solution Service Corporation, which has extensive knowledge of plant sites, to convert chemical plants into Smart Factories. We initially attempted digital twinning of areas that could not be handled by existing automatic control technology by applying machine learning to operating data. Internal movements inside the reactor, however, were so complex that they could not be reproduced using operating data only.
We then adopted another approach for digital twinning the behavior of chemical plant operators. Specifically, we created an AI model through imitation learning using data from past operations under different conditions in processes difficult to control automatically. Demonstration tests comparing the generated AI model with the operator’s operation accuracy confirmed that the content matched with extremely high accuracy, and so in April 2022 the model was commercialized as an “AI plant operation support solution” that realizes AI support for manual operation.The solution has already attracted inquiries from multiple chemical manufacturers, and some have entered the stage of feasibility testing. NTT Communications will continue to improve this solution in a flexible and agile manner while contributing to the sustainable development of the chemical industry and other Japanese industries.


Smart Factory Technology Innovation Center Koji Ito

Koji Ito
Smart Factory Technology Innovation Center

The AI Plant Operation Support Solution provides the reason for AI predicted values, so operators receiving support can confidently exercise control, and it can also be effective in training young operators. Our urgent goal is that—once sufficient accuracy of the knowledge gained by AI is confirmed—we will apply it to plants with all due haste, which could even lead to its automation. And it will be possible to export this skill, which is Japan’s strength, to other countries.
No matter how many improvements are made to the AI Plant Operation Support Solution, as long as some situations continue to depend upon human intervention, the fundamental issues of passing on technology and know-how and securing operators will not be resolved. With this solution, which is a digital twin of human labor, it is difficult to achieve a level of control that exceeds manual operation. We will therefore seek to develop solutions such as those that allow AI to improve plant operation technology on its own in the human-AI dialogue, and will continue to collect, accumulate, and analyze data; connect data and value; and utilize this data for the benefit of society to create Smart Factories.

Note: The content of this article is as of December 2022

Back to Top