Data source authentication for wide-area synchrophasor measurements based on spatial signature extraction and quadratic kernel SVM

• Published in: International journal of electrical power & energy systems Vol. 140; no. C; p. 108083
• Main Authors: Liu, Shengyuan, You, Shutang, Yin, He, Lin, Zhenzhi, Liu, Yilu, Cui, Yi, Yao, Wenxuan, Sundaresh, Lakshmi
• Format: Journal Article
• Language: English
• Published: United Kingdom Elsevier Ltd 01-09-2022; Elsevier
• Subjects: Data source authentication; quadratic kernel support vector machine (QKSVM); Spatial signature extraction; Synchrophasor measurement; wide-area measurement system (WAMS); quadratic kernel support vector machine (QKSVM); Synchrophasor measurement; wide-area measurement system (WAMS); Data source authentication; Spatial signature extraction
• ISSN: 0142-0615; 1879-3517
• DOI: 10.1016/j.ijepes.2022.108083

•The proposed LFC-QKSVM algorithm does not require detailed models and parameters of power systems, making it model-free and more practical in actual applications;•The proposed LFC-QKSVM algorithm is less sensitive to the system size, which can be used for a large number of SMDs (e.g. FDRs) in bulk power systems;•The proposed LFC-QKSVM algorithm can achieve a much higher data source authentication accuracy with a shorter window length using low-reporting measurement data, which means data exceptions can be detected more accurately and timely in practice.
As essential components of the wide-area measurement system (WAMS), phasor measurement units (PMUs), frequency disturbance recorders (FDRs) and universal grid analyzers (UGAs) collect valuable data continuously to reveal the dynamic variations of power systems and to enhance the operators’ situational awareness ability. However, these devices are vulnerable to multiple types of data exception emerging in recent years, such as data source ID mix exception spoofing, substantially threatening system security. To ensure the cyber security of WAMS, this work proposes a new spatial signature extraction method, followed by the quadratic kernel support vector machine (QKSVM)-based algorithm, to authenticate data source in WAMS. First, the load–frequency characteristic (LFC), which can represent the impacts of load variations on frequency, is utilized to extract the spatial signatures of FDRs located in different regions. Then, the quadratic kernel function is employed in the QKSVM-based algorithm to map the signatures into Hilbert space to authenticate the data source more accurately. Finally, case studies in the U.S. Western and Eastern power systems show that the proposed model-free algorithm is less sensitive to system sizes, and can achieve a higher authentication accuracy in a much shorter window length compared with other algorithms.