ABSTRACT: There are large fluctuations in the demand for electricity throughout the day. Utility companies, especially in residential areas, observe a large peak in the demand during late evening hours of the day. In order to maintain such high demands, the companies must run old, inefficient power plants as back-up generators to ensure they can satisfy all demand and retain reliability. To lessen this burden on the power grid, our research considers a model using smart grid technology that redistributes residential electrical usage through autonomous load scheduling. We assume that we know each user’s energy consumption needs, including schedulable loads and base loads. We also consider some users may have future household appliances including plug-in electric vehicles, which require a large amounts of energy to charge, and distributive generators, like solar panels, which when not needed can sell electricity back to the grid. Our model uses a convex optimization algorithm to find an optimal energy consumption schedule for each user in order to reduce the total cost to all users in the system. This algorithm is paired with strategic pricing schedules derived from game theory principles to reduce the peak-to-average-ratio (PAR). Analysis of our simulation shows that our algorithm does reduce the total cost of electrical use by 20%.
Demand response management using game theory for the smart grid
Daniel Goldberg, Lauren Steimle
Washington University in St. Louis
Department of Electrical and Systems Engineering
PhD Supervisor: Zhao Tan
Faculty Supervisor: Arye Nehorai