Privacy-Aware Load Ensemble Control: A Linearly-Solvable MDP Approach

Demand response (DR) programs engage distributed demand-side resources, e.g., controllable residential and commercial loads, in providing ancillary services for electric power systems. Ensembles of these resources can help reducing system load peaks meeting operational limits by adjusting their consumption. To equip utilities or aggregators with adequate decision-support tools ensemble dispatch, we develop a Markov Decision Process (MDP) approach to optimally control ensembles privacy-preserving manner. this end, the concept differential privacy is internalized into MDP routine protect transition probabilities and, thus, DR participants. The proposed also provides trade-off between solution optimality guarantees, analyzed using real-world data from events New York University microgrid York, NY.