引用本文:刘鑫蕊,张修宇,吴泽群,王睿,孙秋野.热网FDI攻击的非侵入式检测方法[J].控制理论与应用,2025,42(7):1265~1274.[点击复制]
LIU Xin-rui,ZHANG Xiu-yu,WU Ze-qun,WANG Rui,SUN Qiu-ye.Non-intrusive detection method of FDI attack in heating network[J].Control Theory & Applications,2025,42(7):1265~1274.[点击复制]
热网FDI攻击的非侵入式检测方法
Non-intrusive detection method of FDI attack in heating network
摘要点击 4218  全文点击 344  投稿时间：2023-08-22  修订日期：2025-04-03
查看全文  查看/发表评论  下载PDF阅读器   HTML
DOI编号  10.7641/CTA.2024.30572
  2025,42(7):1265-1274
中文关键词  FDI  网络攻击  非侵入式检测  灰盒模型  热网
英文关键词  FDI  cyber-attacks  non-intrusive load monitoring  gray box model  heat distributing network
基金项目  国家自然科学基金项目(62173074), 国家重点研发计划项目(2018YFA0702200)资助.
作者单位E-mail
刘鑫蕊* 东北大学 信息科学与工程学院 liuxinrui@ise.neu.edu.cn 
张修宇 东北大学 信息科学与工程学院  
吴泽群 东北大学 信息科学与工程学院  
王睿 东北大学 信息科学与工程学院  
孙秋野 东北大学 信息科学与工程学院  
中文摘要
      针对热网易受网络攻击影响且惯性大的问题, 为提高热网攻击检测的快速性和准确性, 本文首次提出了一 种能够放大攻击带来的状态量偏差的非侵入式在线检测方法, 该方法首先将居住人热行为归纳为黑盒模型, 将房屋 和散热器归纳为白盒模型, 通过白盒与黑盒组成的灰盒模型来计算室内热平衡状态, 其次以室内温度为输入/散失 热量计算的中间量, 放大攻击带来的系统状态量偏差, 最后通过多重匹配状态预测方法进行攻击检测. 为验证所提 方法的有效性, 采用巴厘岛热网模型进行仿真实验, 与传统的检测方法相比, 本文所提方法可以有效放大攻击带来 的状态量偏差, 检测速度和检测率均更高.
英文摘要
      Aiming at the problem that the heat network is susceptible to network attacks and has large inertia, in order to improve the rapidity and accuracy of heat network attack detection, this paper first proposes a non-invasive online detection method that can amplify the state deviation caused by the attack. This method first summarizes the thermal behavior of the occupants as a black box model, and the house and radiator are summarized as a white box model. The gray box model composed of white box and black box is used to calculate the indoor heat balance state. Secondly, the indoor temperature is used as the intermediate amount of input / lost heat calculation to amplify the system state deviation caused by the attack. Finally, the attack detection is carried out by the multi-matching state prediction method. In order to verify the effectiveness of the proposed method, the Bali heating network model is used for simulation experiments. Compared with the traditional detection method, the proposed method can effectively amplify the state deviation caused by the attack, and the detection speed and detection rate are higher.