With the deep penetration of the new generation of information technology in the power system, the power system has gradually evolved into a highly coupled cyber-physical systems (CPS), and false data injection attacks (FDIAs) is the most threatening attack among many power CPS network attacks. Aiming at the problem that the existing knowledge-driven detection process of FDIAs has been in a passive detection state for a long time and ignores the advantages of data-driven active capture of features, an active and passive hybrid detection method for power CPS FDIAs with improved adaptive Kalman filter (AKF) and convolutional neural networks (CNN) is proposed in this paper. First, it analyzes the shortcomings of the traditional AKF algorithm in terms of filtering divergence and calculation speed. The state estimation algorithm based on non-negative positive definite adaptive Kalman filter (NDAKF) is improved, and FDIAs passive detection method with similarity Euclidean distance detection and residual detection as the core is constructed. Then, combined with the advantages of gate recurrent unit (GRU) and CNN in terms of temporal memory and feature expression ability, an active detection method of FDIAs based on GRU-CNN hybrid neural network is designed. Finally, the results of joint knowledge-driven and data-driven parallel detection define a mixed fixed calculation formula, and the active and passive hybrid detection method of FDIAs is established considering the characteristic constraints of the parallel mode. The simulation system example of power CPS FDIAs verifies the effectiveness and accuracy of the method proposed in this paper.
INDEX TERMSPower cyber-physical systems, false data injection attacks, adaptive Kalman filter, gate recurrent unit, convolutional neural networks, active and passive hybrid detection. YANG LI received his Ph.D. degree in Electrical Engineering from