通用逆扰动下的主动配电网合谋攻击防御模型  

Active distribution network collusion attack defense model under universal inverse perturbation

作　　者：刘秀 唐铭 原野 袁新平 王海燕 LIU Xiu;TANG Ming;YUAN Ye;YUAN Xinping;WANG Haiyan(Information Center of Yunnan Power Grid Co.,Ltd.,Kunming 650500,China)

机构地区：[1]云南电网有限责任公司信息中心,云南昆明650500

出　　处：《电子设计工程》2025年第5期125-128,133,共5页Electronic Design Engineering

基　　金：云南电网公司电网管理平台(资产域)二期推广实施配套项目(059300HK42222006)。

摘　　要：为了解决合谋攻击防御下主动配电网受损率高的问题,提出通用逆扰动下的主动配电网合谋攻击防御模型。计算逆扰动深度网络模型的目标函数与通用逆扰动训练的损失函数,建立模型网络深度权重矩阵;利用深度梯度函数对攻击训练样本损失函数进行叠加求导,生成合谋攻击防御数学模型;利用最小网损期望的概念函数求解数学模型,完成主动配电网合谋攻击防御。实验结果表明,所设计模型对2 400个合谋攻击数据包的拦截防御时间都在10 ms以内,数据受损率低于0.5%,与传统模型相比,安全防御能力具有明显优势。In order to solve the problem of high damage rate in active distribution networks under collusion attack defense,a universal anti disturbance active distribution network collusion attack defense model is proposed.Calculate the objective function of the inverse perturbation deep network model and the loss function of the general inverse perturbation training,and establish the deep weight matrix of the model network;Using deep gradient function to superimpose and differentiate the loss function of attack training samples,generating a mathematical model for collusion attack defense;Solve the mathematical model using the concept function of minimum network loss expectation to achieve active distribution network collusion attack defense.The experimental results show that the interception and defense time of the designed model against 2400 collusion attack packets is within 10 ms,and the data damage rate is less than 0.5%.Compared with traditional models,the security defense ability has significant advantages.

关 键 词：通用逆扰动 主动配电网 合谋攻击 防御模型 逆干扰防御 

分 类 号：TN975[电子电信—信号与信息处理]