Abstract
This paper proposes an active low latency attack detection algorithm to improve the cybersecurity of grid-connected photovoltaic (PV) systems. The algorithm is developed by using a hybrid model- and data-driven approach, where the test statistics are formulated by using both the physical model of the PV farm and historical measurements. Unlike most previous detection algorithms that mainly focus on detection accuracy, the proposed algorithm aims at minimizing detection delay while ensuring detection accuracy. The low latency detection algorithm is developed by designing a generalized cumulative sum (CUSUM) detector with a dynamic watermark, which can detect cyberattacks even if the adversary has full knowledge of the system model. To evaluate the performance of the low latency detection algorithm, we propose to measure the stealthiness of cyberattacks by using the Kullback-Leibler (KL) divergence between the pre- and post-attack distributions of the test statistics. Simulation results demonstrate that the proposed algorithm can accurately detect cyberattacks with minimum delays.
Funding Information
  • National Science Foundation (ECCS-1711087)
  • NSF Center for Infrastructure Trustworthiness in Energy Systems

This publication has 32 references indexed in Scilit: