Online Defense of Data Spoofing for Renewable Energy-Based Active Distribution Networks

The diversity of facilities and limitations of communication protocols in active distribution networks (ADNs) make measurements vulnerable to secretive cyber-physical spoofing (CPS) attacks, such as the false data injection attack, impacting applications such as load control, power dispatch, and market planning. To address this critical issue, a novel detection framework is proposed for cyber-physical spoofing defense (CSD), aiming to authenticate data and mitigate the adverse effects of attacks. First, a dynamic local outlier factor (DLOF) is proposed to filter the abnormal measurements and improve the data quality. Then, an enhanced S-transform (EST) is developed to extract the unique fingerprints by mapping the measurements in both the time and frequency domains. Next, a lightweight and highly efficient model is implemented to realize real-time CPS detection after feature extraction. Utilizing the real-world measurements of smart meter from ADNs, multiple comparative experimental results reveal that the proposed detection framework the proposed detection framework achieves a detection accuracy of 96.33%, surpassing the recent advanced method by 18.26% in accuracy. Moreover, the average detection time is 55.62 ms, achieving superior real-time CPS defense.