A game-theoretic approach to decentralized control of heterogeneous load population

This paper investigates the aggregated control problem of a large number of residential responsive loads for various demand response applications. A unified hybrid system model is proposed, capturing the dynamics of both TCLs (Thermostatically Controlled Loads) such as HVACs (Heating, Ventilating and Air Conditioning) and water heaters, and deferrable loads such as washers, dryers, and PHEVs (Plug-in Hybrid Electric Vehicle). Based on the unified hybrid system model, we formulate the aggregated control problem as an optimal control problem, which seeks for an optimal energy usage plan for a population of heterogeneous loads. We then propose a game-theoretic approach to develop a decentralized aggregated control algorithm. Convergence of the proposed algorithm is shown by employing potential game theory. The hybrid system modeling framework and the proposed decentralized aggregated control algorithm are validated through several realistic demand response simulations using GridLAB-D. The simulation results demonstrate that the aggregated power response can accurately track a reference trajectory, effectively reduce the peak power consumption, and efficiently save electricity cost.