A review of nonintrusive load monitoring and its application in commercial building

Abstract

Nonintrusive load monitoring (NILM) has been developed for decades in load monitoring and disaggregation. Methods of digital signal processing and statistical modeling have been applied in detecting state change events, extracting the load signatures, identifying individual device load for disaggregation, and parameterizing the transient of devices for diagnosis. In this paper, we discuss the challenges in general NILM system with methods implemented. We focus on models applied in key steps, which are “event detection” and “load recognition”. Approaches in residential and commercial buildings are evaluated, where these two typical applications have gained lot attentions recently. The challenges in commercial building are explored and studied, and potential solution and research direction are discussed. It shows that unsolved problems and modern intelligent devices require more sophisticated NILM system in the future.