Design of a Smart Controller Agent for Demand-Side Management with Low Payback Effect

Abstract

With high penetration of renewable resources such as wind and solar into conventional electric grid, new challenges are introduced due to the rapid fluctuation on the generation side. Direct load control of thermostatically controlled loads can play a significant role in demand side management (DSM) to cope with the uncertainties and variabilities of the generation. For this purpose, the system operator (SO) requires a reliable forecast of the demand and how much it can be shifted; in order to produce attainable desirable set points to reshape the demand to follow generation side. The focus of this paper is on designing a smart agent that uses a hybrid system of a model-based and a model-free structure to forecast the controllable load and its capacity to be reshaped, and follow the dispatch instructions of the SO, while minimizing the payback effect of the control actions and maintaining customers’ comfort. The main advantages of the proposed system are: 1) real-time model creation; thus, no need for historical data for training, 2) model free controller can automatically adapt to the changes in the system, 3) it can be used as a plug & play component in a DSM program. To evaluate the performance of the proposed controller, a numerical simulator was developed, and the controller was applied over the simulation engine to follow arbitrary desired power profiles. It was observed that the system can follow the dispatch command in less than 5 minutes with a negligible steady state error (less than 5%).