A Secure Power System Distributed State Estimation via a Consensus-Based Mechanism and a Cooperative Trust Management Strategy
Saeed Nasiri, Hossein Seifi, Hamed Delkhosh
Abstract
Future power systems are integrated more and more with digitalized and decentralized platforms in their cyber and physical domains such as industrial IoT (IIoT) technology and distributed energy resources. The first challenge is that these systems need efficient decentralized decision-making approaches for most of their functionalities. The second challenge is the cybersecurity issues where frequent reliable data exchanges among distributed agents are critical and there exist real-world cyber threats. Distributed state estimation (DSE) is a key function for future power systems that is necessary to maintain the system's operating conditions within secure boundaries. However, the aforementioned challenges need to be addressed. To do so, this article <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">first</i> proposes a blockchain framework for secure DSE. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Second</i> , a consensus-based distributed information Kalman filter is proposed, which only needs to communicate the shared states information among the neighbors. The proposed method reduces the computational and communication costs, due to the need for consensus only on the predicted shared states. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Third</i> , a cooperative trust management strategy is developed through which the nodes can vote for each other's trust. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Then</i> , the proposed trust management mechanism is used for anomaly detection, which can accurately and timely detect the abnormal behavior of the nodes and isolate the healthy ones. Numerical simulations on the standard IEEE14-bus test system show the effectiveness of the proposed methods with a much lower mean square error of DSE in both normal and attacked time steps and an F1-score of 0.9945 for anomaly detection.