Resilient Charging Control of Electric Vehicles Against Denial of Service Attacks using Neural Networks
Conference
Podder, AK, Chakrabortty, A, Rahman, MA. (2025). Resilient Charging Control of Electric Vehicles Against Denial of Service Attacks using Neural Networks
. IEEE POWER AND ENERGY SOCIETY GENERAL MEETING 2010, 10.1109/PESGM52009.2025.11225397
Podder, AK, Chakrabortty, A, Rahman, MA. (2025). Resilient Charging Control of Electric Vehicles Against Denial of Service Attacks using Neural Networks
. IEEE POWER AND ENERGY SOCIETY GENERAL MEETING 2010, 10.1109/PESGM52009.2025.11225397
We propose a computationally efficient, data-driven approach using neural networks (NNs) to improve the resilience of power distribution networks integrated with electric vehicle charging stations (EVCSs) during denial of service (DoS) attacks. We build on an existing model-based controller that ensures grid stability by providing optimal charging current setpoints for EVCSs with proper incentivization strategies. The controller requires peer-to-peer communication between the EVCSs. If there is a DoS attack in any communication link, one must re-optimize the setpoints as well as the control gains, which can be computationally intensive and time-consuming. To resolve this, we employ a deep learning architecture using NNs that can predict the setpoints, the control gains, and the corresponding charging incentives, all in near real-time. Another key advantage of this approach is its robustness to model uncertainties, making it inherently more resilient than the model-based solution. Results are validated using the IEEE 33-bus power distribution model.
