Synchrophasor Measurement Prediction using Modified Time-Series Mixer Model under Cyber Attacks

Abstract

Power systems are critical infrastructures that require robust monitoring and control mechanisms to ensure reliability, stability, and efficiency. It utilizes the data from Phasor Measurement Units (PMUs) and other monitoring devices, which are sent over communication links. Hence, they are subjected to cyber attacks leading to missing or corrupted values. This can cause unreliable operation of the power system. To address the issue, this paper proposes a modified Time-Series Mixer model to accurately predict multivariate measurements under continuous unavailability of data. The proposed model utilizes the time-mixing and feature-mixing layers, which help to capture temporal and spatial correlation to predict the measurements. Further, it can be utilized in real-time applications in case of attack on any communication channel/PMU measurements, as the single model predicts all the PMU measurements. The performance of the proposed model is validated using data generated for the IEEE 14 bus system via RTDS. Numerical results validate the effectiveness of the proposed method, as the errors are significantly smaller than those obtained by the LSTM model-based consecutive PMU measurement prediction technique. © 2024 IEEE.