Optimization applied to residential non-intrusive load monitoring

Abstract

This work presents a non-intrusive load monitoring (NILM) method based on mixed-integer linear programming (MILP).NILM are methods for disaggregating measurements from energy meters into information regarding operating appliances.Such information, such as the power consumption and operating state, are valuable for promoting energy savings and predictive maintenance.The proposed technique expands the classical model based on combinatorial optimization (CO).The new formulation handles the problem of ambiguity of similar loads, present in the classical model.Linear constraints are used to efficiently represent load signatures.Additionally, a window-based strategy is proposed to enhance the computational performance of the proposed NILM algorithm.The disaggregation can be made using only active power measurements at low sampling rate, which is already available in commercial smart meters.Other features can be added to the model, if available, such as the reactive power.The performance of the algorithm is evaluated using two test cases from the public dataset AMPds.The sampling rate from the test case is of one sample per minute.Results demonstrate the ability of the proposed method to accurately identify and disaggregate individual energy signatures in a computationally efficient way.