Search

Moving target defense (MTD) in power grids is an emerging defense technique that has gained prominence in the recent past. It aims to solve the long-standing problem of securing the power grid against stealthy attacks. The key idea behind MTD is to introduce periodic/event-triggered controlled changes to the power grid's SCADA network/physical plant, thereby invalidating the knowledge attackers use for crafting stealthy attacks. In this paper, we provide a comprehensive overview of this topic and classify the different ways in which MTD is implemented in power grids. We further introduce the guiding principles behind the design of MTD, key performance metrics, and the associated trade-offs in MTD and identify the future development of MTD for power grid security., Comment: 10 pages, 3 figures, survey

As one of the most sophisticated attacks against power grids, coordinated cyber-physical attacks (CCPAs) damage the power grid's physical infrastructure and use a simultaneous cyber attack to mask its effect. This work proposes a novel approach to detect such attacks and identify the location of the line outages (due to the physical attack). The proposed approach consists of three parts. Firstly, moving target defense (MTD) is applied to expose the physical attack by actively perturbing transmission line reactance via distributed flexible AC transmission system (D-FACTS) devices. MTD invalidates the attackers' knowledge required to mask their physical attack. Secondly, convolution neural networks (CNNs) are applied to localize line outage position from the compromised measurements. Finally, model agnostic meta-learning (MAML) is used to accelerate the training speed of CNN following the topology reconfigurations (due to MTD) and reduce the data/retraining time requirements. Simulations are carried out using IEEE test systems. The experimental results demonstrate that the proposed approach can effectively localize line outages in stealthy CCPAs.

Recently, there has been a surge of interest in adopting deep neural networks (DNNs) for solving the optimal power flow (OPF) problem in power systems. Computing optimal generation dispatch decisions using a trained DNN takes significantly less time when compared to using conventional optimization solvers. However, a major drawback of existing work is that the machine learning models are trained for a specific system topology. Hence, the DNN predictions are only useful as long as the system topology remains unchanged. Changes to the system topology (initiated by the system operator) would require retraining the DNN, which incurs significant training overhead and requires an extensive amount of training data (corresponding to the new system topology). To overcome this drawback, we propose a DNN-based OPF predictor that is trained using a meta-learning (MTL) approach. The key idea behind this approach is to find a common initialization vector that enables fast training for any system topology. The developed OPF-predictor is validated through simulations using benchmark IEEE bus systems. The results show that the MTL approach achieves significant training speeds-ups and requires only a few gradient steps with a few data samples to achieve high OPF prediction accuracy.