CAPED: Context-Aware Powerlet-Based Energy Disaggregation.

Energy disaggregation is the task of decomposing a household’s total electricity consumption into individual appliances, which becomes increasingly important in energy reservation research nowadays. In this paper, we propose a novel algorithm taking the context of disaggregation task into consideration. First, we design a new method to efficiently extract each appliance’s typical consumption patterns, i.e. powerlets. When performing the disaggregation task, we model it as an optimization problem and incorporate context information into the cost function. Experiments on two public datasets have demonstrated the superiority of our algorithm over the state-of-the-art work. The mean improvements of disaggregation accuracy are about 13.7% and 4.8%.