Preferential Cyber Defense for Power Grids

Abstract

To safeguard electric power grids from random failures and cyberattacks, a reinforcement learning approach is developed using linear temporal logic to efficiently allocate limited resources, promising improved defense for critical cyberphysical systems.