Structural-Constrained Methods for the Identification of False Data Injection Attacks in Power Systems

Abstract

Power system functionality is determined on the basis of power system state estimation (PSSE). Thus, corruption of the PSSE may lead to severe consequences, such as disruptions in electricity distribution, maintenance damage, and financial losses. Classical bad data detection (BDD) methods, developed to ensure PSSE reliability, are unable to detect well-designed attacks, named unobservable false data injection (FDI) attacks. In this paper, we develop novel structural-constrained methods for the detection of unobservable FDI attacks and the identification of the attacked buses. The proposed methods are based on formulating structural, sparse constraints on both the attack and the system loads. First, we exploit these constraints in order to compose an appropriate model selection problem. Then, we develop the associated generalized information criterion (GIC) for this problem. However, the GIC method's computational complexity grows exponentially with the network size, which may be prohibitive for large networks. Therefore, based on the proposed structural and sparse constraints, we develop two novel lowcomplexity methods for the practical identification of unobservable FDI attacks: 1) a modification of the state-of-the-art orthogonal matching pursuit (OMP) method; and 2) a method that utilizes the graph Markovian property in power systems, i.e. the second-neighbor relationship between the power data at the system's buses. In order to analyze the performance of the proposed methods, the appropriate oracle and clairvoyant detectors are also derived. The performance of the proposed methods is evaluated on the IEEE-30 bus test case. INDEX TERMSAttack detection and identification, false data injection (FDI) attacks, graph Markovian property, model selection, structural constraints,

State Estimation in Partially Observable Power Systems via Graph Signal Processing Tools

“…. , M , and the TV is not a function of m. 2) For h(L) that satisfy (18) to be a valid graph filter, i.e.…”

“…However, there are various realistic scenarios where the system is partially observable (also named in the literature as unobservable) ([ 1 ] Chpter 4, [ 3 , 4 , 5 ]), that is, the number of sensors is not sufficiently large, or sensors are not well placed in the network. The observability of the system may be compromised due to communication errors, topology changes, sensor failures [ 1 ], malicious attacks [ 6 , 7 , 8 ], and electrical blackouts [ 9 ]. A direct implication of system unobservability is that conventional estimators that assume deterministic states, such as the commonly used weighted least squares (WLS) estimator, can no longer be used since they are inaccurate, inconsistent, and may have large estimation errors even in the absence of noise [ 3 , 10 ].…”

“…Specifically, a graph signal is smooth if the signal values associated with the two end nodes (vertices) of edges with large weights tend to be similar. The smoothness of graph signals has been used for various applications, such as denoising with Laplacian-smoothness regularization [12]- [14], detection of a malfunctioning sensor in sensor networks [1], data classification [1], traffic flow prediction [15], and attack detection in power systems [15]- [19]. Similarly, in GNN-based node classification (see e.g.…”

Estimation of the Admittance Matrix in Power Systems Under Laplacian and Physical Constraints

“…Due to the symmetry of the derivative of the Scaled Augmented Lagrangian function in ( 14) w.r.t. G and B, we can derive the update equation for B(i+1) using a similar procedure as for G in (29). Therefore, under the assumption that H 4 + ρE is non-singular, the update equation for B is given by…”

“…Hence, accurate topology identification and precise line-parameter estimation become crucial for ensuring the efficiency and reliability of the network operations. Additionally, topology identification achieved through admittance matrix estimation can serve as a valuable tool for event detection, such as identifying faults and line outages [28], and detection of potential cyberattacks [9], [29]. Furthermore, with the increasing integration of distributed renewable energy resources, the importance of admittance matrix estimation has risen significantly, making it an indispensable element in addressing the challenges of modern power systems.…”