Glossary · AI engineering and governance
AI-enabled control system
Also known as: AI-based control system
German: KI-gestütztes Steuerungssystem
In automation, an AI-enabled control system is a control system in which one or more functions, such as setpoint optimization, parameter tuning, anomaly handling or decision-making, use machine learning or other AI techniques, alongside or within conventional control logic.
- Industrial AI
- AI
In one sentence
An AI-enabled control system uses machine learning or other AI techniques for functions such as optimization, tuning or decisions.
Example
A kiln control system uses a machine learning model to suggest fuel setpoints, while the PLC enforces fixed limits and the operator confirms changes.
How it applies
- Engineering: A common pattern is to let the AI suggest or optimize within bounds, while deterministic control logic enforces limits and safety functions remain independent of the AI function.
- Compliance: If an AI function performs a safety function in a machine, it may be an Safety component (AI Act) with specific regulatory consequences. Clarify this early with the regulatory and safety teams.
- Operation: Define operator roles: approve, supervise or intervene (Human-in-the-loop (industrial AI), Human-on-the-loop), and provide a fallback mode without the AI function.
- Documentation: Describe what the AI function does, its limits, how operators can recognize and override its actions, and how it behaves when inputs are missing or out of range.
AI-enabled control vs. conventional advanced control
Model predictive control and similar methods also use models, but these are typically derived from physics or system identification and are well understood. AI-based models learned from data are harder to verify, which affects validation and documentation.