An unsupervised approach for disaggregating major loads in small commercial buildings

This paper proposes a new novel way for NonIntrusive Load Monitoring (NILM). The technique can be applied to develop a powerful framework for low-cost power monitoring in buildings, particularly in the small commercial sector. This approach thus solves one of the major challenges in power monitoring and energy management, which has been the development of robust unsupervised learning algorithms that eliminate the need for costly human involvement. A proposal is made about filtering out the major loads in a sequential process from the aggregate power signal. Specifically, exterior lights and Rooftop units are extracted, the latter with the help of an unsupervised clustering method. The method is shown to be computationally compatible with handling a large data set obtained across a large portfolio. Finally, a case is studied for disaggregating major loads obtained from a bank building to demonstrate a basic test case in the real world.