Ensemble-Based Deep Learning Model for Non-Intrusive Load Monitoring

Climate change and environmental concerns are instigating widespread changes in the modern electricity sector due to energy policy initiatives and advances in sustainable technologies. With the deluge of information resulting from the ubiquitous communication and computational capabilities present in all aspects of our society, system operators and consumers have elevated situational awareness and are able to make informed context-based decisions. We capitalize on this information-centric nature of the advanced metering infrastructure (AMI) in the power grid to enable non-intrusive load monitoring for individual consumers with high accuracy. We propose a novel ensemble based deep learning model to disaggregate smart meter readings and identify the operation of individual appliances. We show through comprehensive practical and comparative studies the superior performance of the proposed model.