Neuro-Agentic Control: A Deep Learning-based LLM-Powered Agentic AI Framework for Controlling Security Controls

神经代理控制：一种基于深度学习与大语言模型的代理 AI 框架，用于控制安全防护机制

Abstract: Cyberattacks on operational technology are increasingly causing costly downtime and physical damage, exposing the limitations of traditional rule-based monitoring in industrial IoT environments. While Large Language Models (LLMs) have strong semantic reasoning abilities to assist in decision support, their hallucinatory nature presents unacceptable safety liabilities for closed-loop control.

摘要： 针对运营技术（OT）的网络攻击正日益导致昂贵的停机时间和物理损坏，这暴露了传统基于规则的监控在工业物联网（IIoT）环境中的局限性。虽然大语言模型（LLMs）具备强大的语义推理能力，可辅助决策支持，但其固有的幻觉特性在闭环控制中带来了不可接受的安全隐患。

This paper introduces a neuro-agentic control framework, a novel architecture that couples an LLM-based planner (i.e., such as Gemini 2.5 Flash-Lite) with a pre-trained Time-Series Foundation Model (TimesFM), to achieve physics-grounded autonomous defense.

本文介绍了一种神经代理控制框架（Neuro-Agentic Control Framework），这是一种将基于大语言模型的规划器（例如 Gemini 2.5 Flash-Lite）与预训练的时间序列基础模型（TimesFM）相结合的新型架构，旨在实现基于物理规律的自主防御。

The paper introduces a “Counterfactual Physics Injection” mechanism that simulates the impact of LLM-proposed interventions within the numerical latent space of the foundation model before actuation, while allowing the system to reject hallucinatory or unsafe actions.

本文引入了一种“反事实物理注入”（Counterfactual Physics Injection）机制，该机制在执行操作前，于基础模型的数值潜在空间内模拟大语言模型所提议干预措施的影响，从而允许系统拒绝幻觉或不安全的操作。

Evaluated on an industrial dataset (e.g., the Secure Water Treatment (SWaT)) in the context of stochastic attack scenarios, the framework exhibited better performance compared to LSTM and TCN baselines. The Neuro-Agentic Loop prevented five breaches (33.3%) below the threshold versus LSTM (26.7%) and TCN (13.3%), with zero physically invalid (hallucinated) actions executed.

在随机攻击场景下的工业数据集（如安全水处理系统 SWaT）评估中，该框架表现出优于 LSTM 和 TCN 基准模型的性能。神经代理循环（Neuro-Agentic Loop）成功将低于阈值的违规事件预防率提升至 33.3%（相比之下 LSTM 为 26.7%，TCN 为 13.3%），且执行过程中零次出现物理无效（幻觉）操作。

These results demonstrate the efficacy of using foundation models as deterministic “Sentinels” to safeguard agentic AI in critical infrastructure.

这些结果证明了将基础模型作为确定性“哨兵”来保护关键基础设施中代理 AI 的有效性。