Validation-Gated Multi-Agent Governance for Online Adaptation of Thermal-Hydraulic Surrogate Models under Operating-Regime Shift
arXiv:2606.03321v1 Announce Type: new Abstract: Artificial-intelligence surrogates can support second-by-second thermal-hydraulic forecasting, but models selected and frozen offline may become condition-locked once deployed outside their pretraining envelope. This study develops a guarded continual-adaptation framework for experimental thermal-hydraulic loop data in which role-separated agents - Monitor, Diagnosis, Adaptation, Safety-Auditor, and Orchestrator - diagnose error signatures, prioritize candidate model families, and review promotions, while deterministic champion-challenger gates a
The increasing complexity and autonomy of AI systems necessitate robust validation and governance frameworks, especially in critical real-time applications where model stability and safety are paramount.
This research addresses a fundamental challenge in deploying AI models in dynamic, high-stakes environments by enabling cautious and validated online adaptation, which is crucial for reliability and trust.
The proposed multi-agent governance system allows AI models to continually adapt to changing operating conditions while maintaining safety and performance, moving beyond static, pre-trained deployments.
- · AI model developers
- · High-stakes industries (e.g., energy, manufacturing)
- · Safety-critical autonomous systems
- · Companies relying on brittle, non-adaptive AI systems
- · Traditional static model deployment methodologies
Improved reliability and expanded deployment of AI surrogates in sensitive real-time physical systems.
Accelerated adoption of AI in industrial control and automation due to enhanced trust and operational resilience.
Potential for entirely new classes of self-optimizing, highly adaptive industrial infrastructure managed by AI agents.
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