Controlling cascading failures with cooperative autonomous agents

Cascading failures in electricity networks often result in large blackouts with severe social consequences. A cascading failure typically begins with one or more equipment outages that cause operating constraint violations. When violations persist in a network, they can trigger additional outages which in turn may cause further violations. This paper proposes a method for limiting the social costs of cascading failures by eliminating violations before dependent outages occur. Specifically, our approach places one autonomous software agents at each bus of a power network, each of which is tasked with solving the global control problem with limited data and communication. Each agent builds a simplified model of the network based on locally available data and solves its local problem using model predictive control and cooperation. Through extensive simulations with IEEE test networks, we find that the autonomous agent design meets its goals with limited communication. Experiments also demonstrate that allowing agents to cooperate can vastly improve system performance.