Power disaggregation of domestic smart meter readings using dynamic time warping

Non-intrusive appliance load monitoring (NALM), also called load disaggregation, is a method for isolating the energy consumption of individual appliances from an overall household energy consumption. Traditionally, NALM techniques are based on measuring several electrical parameters at high sampling rates, which increases the meter cost and communications and storage overhead. In this paper, we propose a low-complexity disaggregation method based on Dynamic Time Warping algorithm that uses only active power aggregate smart meter data, captured at a low frequency, for training and disaggregation. Experimental results are provided for three households using data collected during two months, showing which individual appliances were used and when. Average recognition accuracy of 85% was obtained.