Abstract
Large-scale smart energy metering deployment worldwide and integration of smart meters within the smart grid will enable two-way communication between the consumer and energy network, thus ensuring improved response to demand. Energy disaggregation or non-intrusive load monitoring (NILM), namely disaggregation of the total metered electricity consumption down to individual appliances using purely algorithmic tools, is gaining popularity as an added-value that makes the most of meter data. However, NILM remains a challenging problem since NILM is susceptible to sensor noise, unknown load noise, transient spikes, and fluctuations. In this paper, we tackle this problem using novel graph signal processing (GSP) concepts, applied at both, physical signal level via graph-based filtering and data level, via effective semi-supervised GSP-based feature matching. The proposed GSP-based method is generic and can be used to improve results of various event-based NILM approaches. We demonstrate significant improvement in performance using three state-of-the-art NILM methods, both supervised and unsupervised, and real-world active power consumption readings from the REDD and REFIT 1 data sets, sampled at 1 and 8 s, respectively. 1 The REFIT dataset used to generate the results can be accessed via DOI 10.15129/31da3ece-f902-4e95-a093-e0a9536983c4.
Highlights
Integration of smart meters into smart grids, enabled by advanced sensing and communication technologies such as [1] and [2], provides two-way communications between the consumer and energy network to respond in real time to demand [3]
Signal processing alone cannot solve the issue of similarity of appliance loads, that is, very close operational mean power values of two or more appliances. This can be addressed, for example, via an inference approximation method to refine Non-intrusive load monitoring (NILM) results, as done in [12], where Additive Factorial approximate MAP is proposed taking advantage of the additive structure of Factorial HMM (FHMM) and the observation of aggregate power; or with a probabilistic search method as in [8], where simulated annealing is added after the primary graph signal processing (GSP)-based NILM to refine two-state (e.g., ON/OFF) appliance identification by optimizing the difference between the power measurements and the corresponding primary power estimate according to the possible combination of multiple appliances which are switched on simultaneously
WORK This paper addresses the challenging problem of mitigating the effect of measurement noise and unknown loads on load disaggregation (NILM) performance
Summary
Integration of smart meters into smart grids, enabled by advanced sensing and communication technologies such as [1] and [2], provides two-way communications between the consumer and energy network to respond in real time to demand [3]. Current state-of-the-art solutions are susceptible to measurement noise and outliers when dealing with real-world data and do not demonstrate sufficient accuracy [7], [9] One reason for this is the complex nature of the NILM problem with effective solutions requiring both core physical-level signal processing - to process acquired signals reducing jitter, noise, spurious events [8], [27], [34] - and machine learning-based clustering and classification [7]. GSP-based approaches have recently been proposed for tackling the NILM problem, via supervised [8] and unsupervised approaches [27].2 This prior work applied GSP at the data processing stage only, i.e., as a robust classification or clustering tool, without exploiting GSP’s properties as effective physical signal filters [36], which can combat NILM sensitivity to measurement noise and the influence of unknown appliances.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
