Dynamic load altering attack detection in cyber physical power system based on improved Kalman/H∞ co-filtering.

This paper proposes an attack detection scheme for closed-loop Dynamic Load Altering Attacks (DLAA) in Cyber Physical Power Systems (CPPSs). It aims to improve the accuracy of state estimation and attack detection in CPPSs in the presence of D-LAA and noise disturbances. First, a discrete-time CPPSs model subject to D-LAA and unknown-statistics noise is constructed to capture the system dynamics under the influence of cyberattacks and disturbances. Second, a state estimation method based on an improved Kalman/H_∞ co-filter is proposed, in which the multi-fading factor adaptive Kalman filter (MFAKF) is used to handle Gaussian noise with unknown-statistics, and the H_∞ filter is used to enhance robustness against non-Gaussian disturbances. Finally, a detection algorithm based on cosine similarity matching is designed to identify anomalies by calculating the angular deviation between the estimated and measured states. Simulation results show that for the state ω₁ in the IEEE 3-machine 6-bus system, the proposed MFAKF-HF reduces the RMSE by 75% relative to MFAKF and 62% relative to H_∞ filtering, demonstrating the improved accuracy and robustness of the proposed estimation and detection scheme under both attack and noise conditions.