Data‐Driven Modeling, Control, and Tools for Smart Cities

This chapter first presents a method called demand response‐advisor (DR‐Advisor), which acts as a recommender system for the building's facilities manager and provides the power consumption prediction and control actions for meeting the required load curtailment and maximizing the economic reward. Using historical meter and weather data along with set point and schedule information, DR‐Advisor builds a family of interpretable regression trees to learn non parametric data‐driven models for predicting the power consumption of the building. The chapter also presents how data‐driven algorithms can be used for the problems associated with DR and a new algorithm to perform control with regression trees for synthesizing DR strategies. It then describes the MATLAB‐based DR‐Advisor toolbox and provides a comprehensive case study with DR‐Advisor using data from several real buildings. Finally, it summarizes the authors' results and a discussion about future directions.