基于反向鲸鱼-多隐层极限学习机的电网FDIA检测  

FDIA Detection in Power Grid Based on Opposition-Based Whale Optimization Algorithm and Multi-layer Extreme Learning Machine

作　　者：席磊 王艺晓[1] 何苗 程琛 田习龙 XI Lei;WANG Yixiao;HE Miao;CHENG Chen;TIAN Xilong(College of Electrical Engineering and New Energy,China Three Gorges University,Yichang 443002,China;Hubei Provincial Key Laboratory for Operation and Control of Cascaded Hydropower Station(China Three Gorges University),Yichang 443002,China;Substation and Operation Branch of State Grid Jingzhou Electric Power Supply Company,Jingzhou 434000,China)

机构地区：[1]三峡大学电气与新能源学院,湖北宜昌443002 [2]梯级水电站运行与控制湖北省重点实验室(三峡大学),湖北宜昌443002 [3]国网荆州供电公司变电运维分公司,湖北荆州434000

出　　处：《中国电力》2024年第9期20-31,共12页Electric Power

基　　金：国家自然科学基金资助项目(高比例大容量新能源接入的新型电力系统自动发电控制全局类脑分布协同理论与方法研究,52277108)。

摘　　要：针对目前已有的电力信息物理系统虚假数据注入攻击检测方法由于特征表达能力有限,而导致无法精确获取受攻击位置的问题,提出一种基于反向学习鲸鱼优化多隐层极限学习机的虚假数据注入攻击定位检测方法。所提方法不仅将极限学习机拓展为多隐层神经网络,解决其特征表达能力有限的问题,而且引入鲸鱼优化算法对多隐层极限学习机的各隐层神经元个数进行寻优并采用反向学习策略提高其收敛速度和检测精度,以防止随机确定各隐层神经元个数对检测方法的泛化性能和定位检测结果造成影响。通过在不同场景下对IEEE-14和57节点测试系统进行大量实验,验证了所提方法能够通过历史数据自动识别受攻击的系统状态量所对应的精确位置。与其他多种方法相比,所提方法具有更优的精度、召回率和F1值。At present,the existing false data injection attack(FDIA)detection methods for cyber-physical power system can not precisely obtain the location of the attack due to its limited ability of feature expression.Therefore,this paper proposes a FDIA location detection method based on opposition-based learning whale optimization algorithm and multi-layer extreme learning machine(OWOA-ELMML).The proposed method not only extends the extreme learning machine into a multi-layer neural network to solve the problem of its limited ability of feature expression,but also introduces the whale optimization algorithm to optimize the number of neurons of the multi-layer extreme learning machine,and uses the opposition-based learning strategy to improve its convergence speed and detection accuracy so as to prevent the influence of randomly determining the number of neurons in each hidden layer on the generalization performance and location detection results of the detection method.Through a large number of simulation tests on IEEE-14 and 57-bus test systems under different scenarios,it is verified that the proposed method can automatically identify the exact position of the attacked system state through the historical data.Compared with other comparative methods,the proposed method has better precision,recall rate and F1 value.

关 键 词：电力信息物理系统 虚假数据注入攻击 多隐层极限学习机 鲸鱼优化 反向学习 

分 类 号：TM73[电气工程—电力系统及自动化] TP18[自动化与计算机技术—控制理论与控制工程]