AI Control Learning Service

The AI Control Learning Service is Yokogawa's industrial AI platform that packages reinforcement learning AI technology with controllers to enable autonomous plant control. This service represents the world's first successful implementation of AI for autonomous plant control, utilizing Yokogawa's proprietary "Factorial Kernel Dynamic Policy Programming" (FKDPP) algorithm developed jointly with Nara Institute of Science and Technology.

The service addresses common industrial control challenges such as long dead times and large disturbances that are difficult to manage with traditional PID control. Unlike conventional control methods that rely on human intuition and experience, the AI system learns optimal control strategies through minimal trials, making it highly practical for industrial applications.

Key benefits include automated processes that previously required human adjustment, addressing labor shortages in manufacturing. The system improves controllability by suppressing overshoot, reducing equipment load, and shortening settling times, which enhances both productivity and energy efficiency. This contributes to sustainability goals by optimizing quality, yield, and energy consumption simultaneously.

The service operates through a on premise AI control learning tool with annual subscription contracts. Users upload operational data to create simulators, which generate AI control models that can be installed on CPU modules for real-time equipment control. The system supports various industrial applications including temperature, pressure, and flow control across sectors like petrochemicals, electronics, food processing, pharmaceuticals, and water treatment systems.