Boosting Power Systems Resilience Against Counterfeiting Of GPS Attacks Using Deep Learning Approach

Cyber security concerns have been brought up by the expanding usage of wireless technology in power systems, notably regarding GPS counterfeiting attacks (GCAs). Phasor measurement units (PMUs) experience phase angle adjustments as a result of these attacks, which distort GPS data. This work proposes a Deep-learning GPS-Spoofing Counteraction (DLGSC) algorithm that uses PMU data for PMU measurement correction and GCA identification. For signal rectification after the identification of attacks, the proposed technique separately uses an assortment of long short-term memory (LSTM) units and a recurrent neural network (RNN). In contrast with conventional approaches, which either cannot handle concurrent attacks or static, DLGSC addresses these issues by utilizing deep learning methodologies. DLGSC obtains precise results by choosing suitable features for GCA identification. Thereafter total least square-estimation of signal parameters via the rotational invariance technique (TLS-ESPRIT) is used for the modal estimation. The efficacy & resilience of the proposed technique is confirmed and contrasted with other recently discovered methods. Additionally, real probing data, local area data, and two area data from the Western Electricity Coordinating Council (WECC) are used to validate the proposed technique.