Load classification and appliance fingerprinting for residential load monitoring system

Abstract

Previous work on residential load monitoring has attempted to address different requirements including the systematic collection of information about electric power consumption for load research purpose, the provision of a detailed consumption report to facilitate energy conservation practices and the monitoring of critical loads for fault diagnostics.

This work focuses on developing methods for appliance fingerprinting that is foreseen to be an integral part of an automatic residential load monitoring system. Various approaches outlined in previous research form the basis for the concepts developed in this thesis. In addition, an extensive series of measurement work was performed on several household appliances in order to acquire the necessary operation data for building the technique and also to explore the extent up to which residential loads can be categorized into distinct groups.

The fingerprinting process proposed in this work employs three main phases: feature extraction of electrical attributes, event detection and pattern recognition. Test results obtained at different stages of the work using the measurement data are also discussed in detail.

Such studies are necessary to enable utilities to manage their networks reliably and efficiently, and also to encourage the active participation of consumers in energy conservation programs.