An Effective CPT-Based Nonintrusive Load Monitoring for Cognitive Meters

Abstract

Nonintrusive Load Monitoring (NILM) is the process to disaggregate the power consumption in appliances using a single meter installed on the residence electric entrance. We introduce a novel load modeling approach for power consumption disaggregation using the equivalent circuit concept from the Conservative Power Theory (CPT). CPT can represent an appliance or an operation stage using an equivalent circuit based on load parameters. This method allows obtaining load disaggregation models efficiently, easily, and fast, without requiring to collect and label plenty of samples. To validate it, we have (i) used a meter to capture signals from the aggregate consumption; (ii) extracted the features from the CPT equivalent circuit parameters for appliance recognition in a load event; (iii) created and validated a classifier constructed through equivalent circuits of the appliances, and (iv) tested the classifier in NILM real-world case studies. Experiments from different appliances achieved promising results, indicating effective load disaggregation capacity and reinforcing CPT as groundwork for load characterization towards an efficient load disaggregation.