An Experimental Design for Identification of Multiple Load Appliances

This paper presents an experimental design for building an efficient energy disaggregation system through multi-label classification approach. The proposed system requires a single point measurement of common electrical parameter data at aggregate electric circuit to identify the operating status of multiple appliances. Some multi-label classification algorithms were evaluated to select the best one that could provide the optimal classification performance. The results showed that the Binary Relevance (BR) and Decision Tree (C4.5) as a pair of the multi-label classifier and based classifier could identify the appliances by effective processing time with an accuracy of 99% for water heater and air conditioner and above 86% for lightings and other utilities.