A Training-free Non-intrusive Load Monitoring Approach for High-frequency Measurements based on Graph Signal Processing

Abstract

As an enabling technology for Demand Side Management, non-intrusive load monitoring (NILM) offers users appliance-level energy usage feedback via analysis of the total energy measurements collected by a single metering device. Compared to NILM on low-rate measurements, more transient characteristics of individual appliances can be extracted in NILM high-rate measurements. Since appliances’ state transitions are less likely to overlap on high-rate measurements, high-rate NILM can be treated as a classification problem. In this paper, we propose a NILM classifier based on the emerging graph signal processing (GSP) concepts, which has shown promising advantage in dealing with high dimensional and complex data in signal denoising, clustering and classification, etc. Besides active power, current harmonics, reactive power and V-I trajectory are exploited as load features in the proposed method. The method is compared with the state-of-the-art NILM methods based on measurements for two houses from publicly-accessible REDD dataset in various evaluation metrics and outperforms benchmarks.