机器学习缓解的广域阻尼控制系统异常检测

为了建立攻击弹性，以抵抗对测量信号和控制信号段的隐蔽网络攻击，提出了一种基于机器学习缓解策略的广域阻尼控制系统异常检测方法。首先提出基于信号熵的特征提取，从而提高机器学习模型的训练检测精度和鲁棒性。然后提出一种基于电力系统运行条件和网络攻击事件的组合数据集生成方法，以便用于任何大规模电网模型。引入的缓解模块能够调谐系统信号，并同时在测量和控制信号上进行攻击检测。在2区域4机电力系统的测试环境下对本文方法的性能进行了评估，结果表明本文方法能够实现高精度的异常检测。

Anomaly Detection of Wide Area Damping Control System Based on Machine Learning Mitigation

In order to establish attack resilience to resist covert network attacks on measurement signal and control signal segments, an anomaly detection method for wide area damping control system based on machine learning mitigation strategy was proposed. Firstly, feature extraction based on signal entropy was introduced to improve the detection accuracy and robustness of machine learning model. Then, a combined data set generation method based on power system operating conditions and network attack events was proposed for any large-scale power grid model. The introduced mitigation module could tune the system signal and detect the attack on the measurement and control signals at the same time. The performance of this method was evaluated in the test environment of 2-area 4-machine power system. The results show that this method can achieve high-precision anomaly detection.