An Online Learning Framework for Targeting Demand Response Customers

We study the optimal design of demand response and peak-time rebate programs in the electricity sector. Load-serving entities (LSEs) frequently use demand response or peak-time rebate programs to reduce consumption during peak hours. However, LSEs have imperfect information, so they typically estimate a customer’s counterfactual energy consumption—often referred to as ‘baseline’— and offer compensations for reductions below this threshold. Due to the inherent variability and uncertainty of demand, this approach may lead to generous incentives to customers who make no effort to reduce their demand, and underpayment to those who do. We show that the estimated baseline is not necessarily the optimal threshold for calculating rebates in demand response programs. The LSE needs to learn about customers’ demand patterns and also customers’ propensity for reductions over time. We bring in tools and methodologies from online learning to formulate and solve the LSE’s optimization problem in designing demand response program. By appropriately tuning the parameters, the LSE can use this framework to balance multiple objectives to determine the appropriate rebate threshold for customers. Using the data from a ToU pricing experiment in London, we show that compared to a conventional demand response program, our approach significantly improves the targeting of demand response payments.