Forecasting Refrigerators Consumption to Support their Aggregated Participation in Demand Response

Demand response programs have become very relevant. However, one of the important facts to have a reliable DR program is the creation of a clear and trustable perspective of the load consumption during the upcoming time periods. On another hand, the increment of the energy-based systems and different energy consuming appliances in the last decades results in larger daily energy consumption which creates the unpredictability of the energy demand. This variety of consumption profiles requires not only consideration of the total consumption of each consumer, but more detailed and focused studies on each type of energy consuming devices. Therefore, this paper proposes a system containing a combination of different forecasting and clustering algorithms to predict the power consumption of several refrigerators and aggregate them into certain groups based on the characteristics of their consumption profiles. Sequentially, the obtained results will be aggregated and serve as basis in order to schedule refrigerators in the context of a demand response program. In the case-study, 20000 refrigerators are considered.