Time-frequency based cyber security defense of wide-area control system for fast frequency reserve
Time-frequency based cyber security defense of wide-area control system for fast frequency reserve
复制标题
DOI：
10.1016/j.ijepes.2021.107151
复制
发表时间：
2021-11
影响因子：
5.2
通讯作者：
W. Qiu;Kaiqi Sun;Wenxuan Yao;Shutang You;H. Yin;Xiaoyang Ma;Yilu Liu
中科院分区：
文献类型：
--
作者：
W. Qiu;Kaiqi Sun;Wenxuan Yao;Shutang You;H. Yin;Xiaoyang Ma;Yilu Liu
全球电力系统正在从传统的化石燃料能源过渡到可再生能源，因为它们具有环境效益。可再生能源的日益普及对电力系统运行提出了挑战。响应储备的效率和充足性对于具有高比例可再生能源的电力系统变得越来越重要。快速频率储备（FFR），特别是基于广域监测系统（WAMS）的快速频率储备是一种有前途的有效解决方案，以确保和提高电力系统的稳定性。然而，网络安全已成为基于WAMS的FFR系统的新挑战。由于FFR的快速功率可控性要求，对FFR控制系统的网络攻击可能威胁电力系统运行的安全。为了解决这个问题，提出了一种基于时间-频率的网络安全防御框架，以检测基于WAMS的FFR控制系统中同步相量数据的网络欺骗。本文首先介绍了连续小波变换（CWT）分解欺骗信号。然后，提出了双频尺度卷积神经网络（DSCNN）从两个频率尺度识别时频域矩阵。结合CWT和DSCNN，进一步提出了一种识别框架CWTs-DSCNN，用于检测基于WAMS的FFR系统中的欺骗攻击。使用FNET/GridEye的实际数据进行多个实验，以验证该框架在保护基于WAMS的FFR系统中的有效性。
Global power systems are transiting from conventional fossil fuel energy to renewable energies due to their environmental benefits. The increasing penetration of renewable energies presents challenges for power system operation. The efficiency and sufficiency of responsive reserves have become increasingly important for power systems with a high proportion of renewable energies. The Fast Frequency Reserve (FFR), especially the Wide-area Monitoring System (WAMS)-based FFR, is a promising and effective solution to secure and enhance the stability of power systems. However, cyber security has become a new challenge for the WAMS-based FFR system. Cyber attacks on the FFR control system may threaten the safety of power system operation due to the rapid power controllability requirement of FFR. To address this problem, a time-frequency based cyber security defense framework is proposed to detect the cyber spoofing of synchrophasor data in WAMS-based FFR control systems. This paper first introduces the Continuous Wavelet Transforms (CWTs) to decompose spoofing signals. Then, the Dual-frequency Scale Convolutional Neural Networks (DSCNN) is proposed to identify the time-frequency domains matrix from two frequency scales. Integrating CWTs and DSCNN, an identification framework called CWTs-DSCNN is further proposed to detect the spoofing attacks in the WAMS-based FFR system. Multiple experiments using the actual data from FNET/GridEye are performed to verify the effectiveness of the framework in securing WAMS-based FFR systems.