Multiarea Probabilistic Forecasting-Aided Interval State Estimation for FDIA Identification in Power Distribution Networks

Power distribution networks are evolving toward a modernized cyber-physical system that is vulnerable to mounting cybersecurity threats brought by false data injection attacks (FDIAs). To address the issue, this article innovatively proposes a multiarea probabilistic forecasting-aided interval state estimation (MPF-ISE) framework for FDIA identification in power distribution networks. The framework devises a novel probabilistic forecasting (PF) approach to achieve interval pseudo measurement modeling. A nonlinear programming-based training algorithm is formulated to minimize the interval width of forecasting error quantiles, which are analytically approximated by Cornish-Fisher expansion. Next, real-time interval measurements are constructed based on the unknown-but-bounded theory. By implementing interval measurement conversion, the local PF-ISE model is converted into rectangular coordinate forms and iteratively solved using a modified Krawczyk operator. Then, the local model is extended to a multiarea form and solved by considering the information exchange between adjacent subareas. Finally, the solution of MPF-ISE, which takes into account measurement uncertainties and line parameter variations, is regarded as the normal operating level, thus forming the proposed FDIA identification scheme. Case studies on the modified IEEE 123-node test feeder demonstrate the superiority of the proposed PF and the effectiveness of MPF-ISE in identifying FDIAs compared with existing ones.