Certifiable, Attack-resilient Submodular Control Framework for Smart Grid Stability

The smart grid is a large-scale, societal-level hybrid cyber-physical system with tight coupling between cyber and physical components.  Ensuring availability and reliability of power requires maintaining stability of the power grid even as increasing demand and uncertain renewable power sources push the power system close to its operation limit.  In addition, the cyber-enabled grid has multiple entry points, leaving it highly susceptible to cyber attacks by malicious adversaries.  Currently, however, developing scalable, certifiable, bound-achieving, and attack-resilient control methodologies for power system stability in the context of Science, Technology, and Engineering of CPS is an open and vital research area.

We will research and develop a submodular framework that will provide control methodologies that are scalable, certifiable, and attack-resilient for the following power system problems: (i) voltage stability, (ii) small-signal stability, and (iii) transient stability.  Submodularity is a diminishing returns property that enables development of efficient algorithms with provable optimality guarantees.  Our main insight is that the smart grid stability problems have inherent physical invariants that exhibit submodularity in terms of control variables.  When submodular structures are exhibited by the physical dynamics, scalable algorithms can be developed to select control actions with certifiable stability guarantees, thus eliminating the computationally expensive current practice of computing control actions and verifying stability through simulation.  Another key observation is that submodular structures remain intact under cyber attacks.