Another perspective on NILM systems for informed energy consumer behavior
Authors
Abstract
Environmental and sustainability issues motivate consumers to reduce their energy consumption, namely the electricity used in their households. When it comes to altering energy behavior it is crucial for users to understand how much electricity single appliances use. However, feedback in pure numbers is rather technical and not appropriate for most consumers. In fact, several studies point out the importance of adequate feedback on electrical consumption of a household to lead to an effective alteration of behavior in order to achieve energy reduction.In related literature it is shown that individual appliance’s loads can be computed from sampling whole-house power consumption signal at one single point. This has been known as Non-Intrusive Load Monitoring (NILM). The acquired signal must then be processed to extract the relevant electrical features, namely for the identification of the different appliances used and its respective consumption.
We propose to formulate the issue of individual load separation as a single channel source separation problem. The on-going research takes the electrical signal disaggregation task considering it as a source separation problem thus overcoming technicalities by using single current state-of-the-art sensing devices. The results so far and focusing on a framework based on non-negative factorization methods for source separation are very encouraging for a final electrical disambiguation system that enables users to identify appliances consumptions in their residencies. It is then possible to help users to better assess their electricity consumption and draw effective measures to save electricity by using non-invasive and ubiquitous systems provided at virtually low-cost market prices.