Machine Learning-Aided Enhancement of Power Grid Resilience to Electromagnetic Pulse Strikes

The electromagnetic pulse (EMP) disturbances could potentially invoke irreversible damages to a set of electric equipment and the power grid at large, which would result in severe power outages. With the EMP attacks recognized as an emerging concern for the power transmission systems, it calls for investigating effective strategies that enable prevention and alleviation of such malicious threats. This paper proposes a new artificial intelligence (AI)-aided mitigation strategy against EMP strikes via adjusting the connection status of shunt capacitors and transformers in the system. A Convolutional Neural Network (CNN) with Autoencoder (AE) deep learning framework is presented for an efficient strategy selection under a large number of EMP strike scenarios simulated in the PowerWorld simulation environment. The numerical studies on the IEEE 30-bus test system demonstrate that the proposed mitigation scheme with the embedded deep learning framework can intelligently provide actionable mitigation solutions in response to EMP scenarios even under interference disturbances (e.g., noise, missing data, etc.).