PulseAugur
实时 11:01:39
English(EN) Validation-Gated Multi-Agent Governance for Online Adaptation of Thermal-Hydraulic Surrogate Models under Operating-Regime Shift

多智能体系统自适应热工水力AI模型

研究人员开发了一种新颖的多智能体治理框架,旨在实现热工水力代理模型的在线自适应。该系统使用独立的智能体进行监控、诊断、自适应、安全审计和编排来管理模型更新。与静态部署相比,该框架在预测准确性方面提高了 19.0%,平均绝对误差为 5.72,并将警告超限减少了 35.8%。这种方法允许对代理模型进行可审计的演进,同时保持对模型部署的控制。 AI

影响 为关键系统的自适应AI模型部署引入了一个新颖的框架,有可能提高预测准确性和可靠性。

排序理由 该集群包含一篇详细介绍新AI方法的学术论文。

在 arXiv cs.MA (Multiagent) 阅读 →

AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →

报道来源 [2]

  1. arXiv cs.LG TIER_1 English(EN) · Doyeong Lim, Seungyoon Lee, In Cheol Bang ·

    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 …

  2. arXiv cs.MA (Multiagent) TIER_1 English(EN) · In Cheol Bang ·

    Validation-Gated Multi-Agent Governance for Online Adaptation of Thermal-Hydraulic Surrogate Models under Operating-Regime Shift

    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 exp…