arXiv:2606.28011v1 Announce Type: cross Abstract: We propose an agentic Large Language Model (LLM) framework for active Fault-Tolerant Control (FTC) that transforms fault detection outputs into constraint-aware recovery actions grounded in plant-specific knowledge. The approach couples (i) a multi-agent workflow that decomposes operator duties into monitoring, planning, action synthesis, simulation, validation, and reprompting; (ii) a Digital Process Plant Twin (DPPT) that exposes plant data, models, and a simulation service for pre-execution testing; and (iii) a Graph Retrieval-Augmented Gene
Advances in LLMs and AI agent frameworks are enabling increasingly sophisticated applications in control systems, moving beyond theoretical fault detection to active remediation.
This development allows critical infrastructure and industrial processes to become more resilient and autonomous, reducing human intervention in managing complex failures.
Control systems can now leverage LLMs for not just identifying faults but also for planning, simulating, and executing corrective actions based on contextual knowledge.
- · Industrial automation sector
- · Critical infrastructure operators
- · AI agent developers
- · Manufacturers of complex machinery
- · Traditional fault detection system providers
- · Manual control system operators (long-term)
- · Industries resistant to AI integration
Industrial systems will exhibit higher uptime and reduced operational costs due to automated fault recovery.
The demand for skilled human operators will shift from routine monitoring to higher-level supervision and system design.
Enhanced autonomy in critical systems could accelerate the development of fully automated factories and smart cities, but also raise new questions about safety and liability.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG