Modeling and Analysis of Baseline Manipulation in Demand Response Programs

Abstract

Baseline methods are used in demand response (DR) programs to estimate customers’ intrinsic load so as to reward them properly. While the accuracy of baseline methods has drawn considerable attention, the strategic behavior regarding baseline manipulation has not been well explored in the literature. In this paper, we formulate the customer’s payoff-maximizing problem as a Markov decision process (MDP). Several structural results have been established, including the characterization of underconsumption on event days and overconsumption on non-event days. We investigate the approximation of baseline methods to understand how the method parameters and the consumption statistics would affect the strategic behavior. Moreover, we develop a rollout algorithm, based on approximate dynamic programming, to solve the MDP efficiently. Finally, the proposed methodology is illustrated through case studies, which shed light on the analysis and design of baseline methods.