Improving electrical grid state estimation accuracy under cyber-attacks using median filtering

The purpose of the work is application of median filtering for increasing the accuracy of power system state estimation (SE) under cyber-attacks. Median filtering (MF) is a nonlinear signal processing technique widely used in digital image processing. Median filtering is performed by letting a window move over the points of a sequence and replacing the value at the window center with the median of the original values within the window. The advantage of the proposed method is the ability of median filtering to discard the values with low accuracy, thus avoiding their impact on general state estimation of power systems. It should be mentioned that the method is easy to implement. MF advantage foundation was performed by modeling of an electrical grid section in the software package PSCAD. The measurements of currents and voltages at each node of the power system were simulated using phasor measurement units. State estimation was performed on the basis of the obtained measurements by the weighted least-squares method with additional one-dimensional and two-dimensional median filtering. The results of state estimation by the specified methods have shown inefficiency of the weighed least-squares method under cyber-attacks on communication channels, since one rough error in the measurement array “spoils” the overall state estimation results, whereas median filtering provides high accuracy. Unlike the weighed least-squares method used for state estimation the proposed method of median filtering allows to reduce the impact of cyber-attacks on the power system. Thus, median filtering is considered to be a promising tool for solving state estimation tasks, the use of which will lead to higher reliability of electrical grids in case of cyber-attacks.

smart grids, cyber security, cyber threat, cyberattack, weighted least-squares method, median filtering