智能电网检测器通常被视为黑箱,仅在干净、时间上打乱的数据集上进行验证,这可能导致过度乐观的性能声称。本研究提出了一种基于可解释、时间窗口融合随机森林的多测试平台评估研究,当评估协议不一致时采用悲观结果。 (i) 在公共配对的智能电网数据集上,打乱的交叉验证结果乐观:宏观F1值从0.96降至0.88,±0.06,在滚动原点时间评估下。一个无需拓扑结构的图卷积网络(GCN)证实了该轻量模型在这些分割上的竞争力。 (ii) 自适应且自适应联合特征空间攻击,尽管在真实电网操控中不可行,导致宏观F1值分别降至0.67和0.34。
推荐理由:来自arXiv AI的《Interpretable Cyber-Physical Attack Detection in Smart Grids: A Leakage-Controlled Multi-Testbed Evaluation of Fusion, Robustness, and Observability Limits》。重点看评测与基准、文化创意。摘要提到:这条英文动态主要涉及评测与基准、文化创意。原文要点:Cyber-physical smart-grid detectors are typically treated as black boxes and validated exclusively on clean, temporally shuffled datasets, which risk overly optimistic performance claims. This work presents a multi-testbed evaluation study built around an interpretable, time-windowed fusion Random Forest, adopting the pessimistic result whenever evaluation protocols disagree. (i) On