基于ARO-MKELM的微电网攻击检测

Attack Detection of Microgrid Based on ARO-MKELM

智能电网的复杂性和开放性使其在信息交换时更易受到网络攻击的威胁。目前大多数检测方法只关注检测攻击的存在性,不能确定受到攻击的分布式电源的具体位置,导致无法快速将被攻击的分布式电源隔离,继而造成严重的损失。提出一种基于人工兔群优化算法优化多核极限学习机的交流微电网虚假数据注入攻击检测方法。在传统极限学习机中引入组合核函数以提升检测模型的学习能力和泛化能力,并采用具有强全局搜索能力的人工兔群优化算法优化多核极限学习机的核函数参数及正则化系数,进一步提升检测模型的检测精度。利用非训练样本内幅值为55和95的阶跃攻击信号进行仿真验证,检测准确率范围分别达到了(93.44~94.64)%和(98.11~99.23)%,与其他检测模型进行对比分析,验证了所提方法的优越性。

The complexity and openness of smart grids make them more vulnerable to the threat of cyber attacks when exchanging information.Most of the current detection methods only focus on detecting the existence of the attack and cannot determine the specific location of the attacked distributed power supply,which results in the inability to quickly isolate the attacked distributed power supply and subsequently cause serious damage.To propose a false data injection attack detection method for AC microgrids based on the artificial rabbit swarm optimization algorithm optimizing a multi-core limit learning machine.A combined kernel function is introduced into the traditional extreme learning machine to improve the learning ability and generalization ability of the detection model,and the artificial rabbit pack optimization algorithm with strong global search ability is used to optimize the kernel function parameters and regularization coefficients of the multi-core extreme learning machine to further improve the detection accuracy of the detection model.Simulation verification is carried out using step attack signals with amplitudes of 55 and 95 within the non-training samples,and the detection accuracy ranges from(93.44~94.64)%and(98.11~99.23)%,respectively,and comparative analyses are carried out with other detection models to verify the superiority of the proposed method.