Immunity Toward Data-Injection Attacks Using Multisensor Track Fusion-Based Model Prediction

Utilization of synchrophasor measurements for wide-area monitoring applications enables system operators to acquire real-time grid information. However, intentional injections of false synchrophasor measurements can potentially lead to inappropriate control actions, jeopardizing the security, and reliability of power transmission networks. To resolve this issue, a multisensor track-level fusion-based model prediction (TFMP) has been proposed. It has been demonstrated on a mature wide-area monitoring application, which detect electromechanical oscillations. In this paper, to extract the initial correlation information about attacked oscillation parameters, Kalman-like particle filter (KLPF)-based smoother has been used at each monitoring node. To reduce its computational burden, the KLPF-based smoother is diagonalized into subsystems. The scheme is further supported by the characteristics of moving horizon estimates for handling continuous load fluctuations and perturbations caused by data injections in power grids. Performance evaluations are conducted using different data-injection scenarios in the IEEE New England 39 Bus system. Results show the proposed TFMP accurately extracted oscillatory parameters from the contaminated measurements in the presence of multiple system disturbances and random data injections.