Targeted Messaging for Appliance-based Demand Response

DR programs could be greatly improved through a targeted personalized approach that could decrease the disruption of consumers’ daily lives and increase their engagement and responsiveness. This paper contributes to this direction by introducing an appliance-based DR strategy for peak shaving. This strategy is based on a selection mechanism for the aggregator implementing the DR program, aiming to reduce the size of the consumer group receiving alerts for a DR request. In this context, for each DR request only the most appropriate subset of consumers is alerted, according to the consumption profile of their appliances. The proposed approach is evaluated on the REFIT dataset, where it achieves significantly higher precision and recall compared to random subset selection (0.795 vs 0.503 recall for a subset with 50% of the total households). Directions for future research are also discussed.