Multi-objective Optimization for Power Load Recommendation Considering User’s Comfort

This paper proposes a recommendation system for load shifting of energy consumption for residential consumers. The main goal is to provide to customer a set of energy consumption strategies, which would span from maximum cost saving strategy, to maximum comfort preserving strategy. The discomfort of user caused by load shifting is expressed here as a Euclidean distance between recommended and forecasted consumption. Recommendation is formulated as a multi-objective optimization problem, solved by NSGA-II (non-dominated sorting genetic algorithm II). Evaluation of proposed method is carried out on data from Pecan Street [1], which were preprocessed and aggregated, to form a typical consumption and photovoltaic (PV) generation course for winter and summer day. Albeit no batteries are present in original dataset, we also consider employing the batteries for storing PV generated spare power, with simple heuristics to control charging and discharging the batteries.