Online Demand Response for End-User Loads

Smart grids as digitalized electricity networks can provide many new capabilities such as managing an enormous number of distributed energy resources, supporting large quantities of renewable energy productions even in small-scales, as well as enabling demand side to participate more actively in demand response (DR) programs. In the depth of digital communication capabilities, alternative decision-making tools are needed for providing adequate solutions to satisfy the involved customers with the new reality: decisions have to be made fast (online) and with the scarce information about the future. However, the state-of-the-art on DR has been providing decision-making tools based on conventional optimization framework that are carried offline, while the real-time nature of most of DR programs requires online optimization approaches. In this regard, we present an online DR model for an end-user load that receives price information on real time and decides about the next action in a completely online fashion. Then, we present an algorithm based on the gradient descent method for solving the proposed DR model. The theoretical model and its applicability are presented and verified using numerical simulations. The results demonstrate the ability to reach considerable profits in a simple and easy-to-implement procedure with limited exogenous data and no information about future random prices.