Evaluation Of Machine Learning Algorithms For Non-Intrusive Load Monitoring And Wireless Control

Authors

  • Muhammad Zubair Salahuddin Chishti , Ir. Jacqueline Lukose

Abstract

Purpose- This research focuses on the evaluation of 15 Artificial Intelligence (AI) algorithms in
Non-Intrusive Load Monitoring System (NILMS). It reveals that among other algorithms Quadratic State
Vector Machine (SVM) followed by Ensemble Bagged Trees algorithm and Artificial Neural Network (ANN)
provides better recognition rate with prediction accuracy of 97.6%, 97.5% and 96.9% respectively. At total 6
loads including 2 identical incandescent lamps, an iron, a blender, LED floodlight and a laptop were chosen
as test samples to represent the wide variety of load types. Th

Published

2020-01-31

Issue

Section

Articles