Mixed-Integer Nonlinear Programming for State-Based Non-Intrusive Load Monitoring

Energy disaggregation, known in the literature as Non-Intrusive Load Monitoring (NILM), is task of inferring energy consumption each appliance given aggregate signal recorded by a single smart meter. In this paper, we propose novel two-stage optimization-based approach for disaggregation. first phase, small training set consisting disaggregated power profiles used to estimate parameters and states solving mixed integer programming problem. Once model are estimated, disaggregation problem formulated constrained binary quadratic optimization We incorporate penalty terms that exploit prior knowledge on how traces generated, appliance-specific constraints characterizing signature different types appliances operating simultaneously. Our compared with existing algorithms both synthetic dataset three real-world datasets. The proposed formulation computationally efficient, able disambiguate loads similar patterns, successfully reconstruct signatures despite presence unmetered devices, thus overcoming main drawbacks methods available literature.