Defending Against Adversarial Attacks by Energy Storage Facility

Abstract

Adversarial attacks on data-driven algorithms applied in power system will be a new type of threat on grid security. Literature has demonstrated the adversarial attack on deep-neural network can significantly misleading the load forecast of a power system. However, it is unclear how the new type of attack impact on the operation of grid system. In this research, we point out that adversarial attacks on the algorithms are distinct from traditional attacks. Rather than causing system collapse, the adversarial attacks camouflaged as normal forecast bias and thus is able to continuously cause the economic loss in a system. We manifest that the adversarial algorithm attack induces a significant cost-increase risk which will be exacerbated by the growing penetration of intermittent renewable energy. In Texas, a 5% adversarial attack can increase the total generation cost by 17% in a quarter, which account for around <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\$ 2\times 10^{7}$</tex> . When wind-energy penetration increases to over 40%, the 5% adversarial attack will inflate the generation cost by 23%. Our research discovers a novel approach of defending against the adversarial attack: investing on energy-storage system. All current literature focuses on developing algorithm to defending against adversarial attack. We are the first research revealing the capability of using facility in physical system to defending against the adversarial algorithm attack in a system of Internet of Thing, such as smart grid system.