Optimization Algorithms for Catching Data Manipulators in Power System Estimation Loops

Abstract

In this paper, we develop a set of algorithms that can detect the identities of malicious data manipulators in distributed optimization loops for estimating oscillation modes in large-power system models. The estimation is posed in terms of a consensus problem among multiple local estimators that jointly solve for the characteristic polynomial of the network model. If any of these local estimates is compromised by a malicious attacker, resulting in an incorrect value of the consensus variable, then the entire estimation loop can be destabilized. We present four iterative algorithms by which this instability can be quickly detected, and the identities of the compromised estimators can be revealed. The algorithms are solely based on the computed values of the estimates, and do not need any information about the model of the power system. Both large and covert attacks are considered. The results are illustrated using simulations of the IEEE 68-bus power system model.