Non-Intrusive Load Monitoring using Artificial Neural Network

Muhammad Zubair Salahuddin Chishti 1


The project has been taken to perform research on Non-Intrusive Load Monitoring (NILM) using Artificial Neural Network (ANN) algorithm. For training ANN, Backpropagation Lebvenberg Marquardt (LM) algorithm have been employed instead of Backpropagation Gradient Descent (GD) due to its prediction accuracy of 87.2% for sixteen different switching combinations of 4 loads. The developed system uses interactive GUI to monitor real-time energy consumed by the loads every second. The load disaggregation technique for detection of sixteen possible occurrence of four load's events has been suggested as a suitable way of implementing NILM. The achieved prediction accuracy for the research gives, practically, a good fit countering the adverse effect of overfitting experienced in achieving high prediction accuracy in AI algorithms.

Type of Paper: Conceptual

Keywords: Non-Intrusive; Artificial Intelligence; Simultaneous Load's Events; Internet of Things

Global Journal of Engineering & Technology Review Conferences 2018-19:
Meet Inspiring Speakers and Experts at our 1000+ Global Annual Meetings