False Data Injection Attack Identification Approach Using Generalized Correntropy Loss Unscented Kalman Filter

The concept of False Data Injection Attack (FDIA) is a prevalent cyber threat to the modern smart grids that uncertainly manipulates measurement data and evades the bad data detection system of the grid. Such a critial scenario is responsible for reducing the state estimation accuracy. The present work investigates the impact of FIDA on non-Gaussian measurement noise by implementing the Generalized Correntropy Loss (GCL)-supported Unscented Kalman Filter (UKF), often known as GCL-UKF. The Mean Square Error (MSE) Loss of UKF is replaced by GCL to mitigate the non-Gaussian interference while maintaining the robustness of UKF to deal with the large model nonlinearities. Extensive simulation results are conducted on IEEE 9-bus, 14-bus, and 30 -bus systems to showcase the efficacy of the GCL-UKF algorithm in comparison with the benchmark methods. The suggested technique not only enhances the state estimation accuracy but also demonstrates promising performance in real-time state estimation applications.