Non-Intrusive Load Monitoring for a Smart House

PhD Project

Project Overview

In this project, we will investigate the use of machine learning techniques for identifying electrical loads (appliances) in a smart house environment. The main approaches explored will include unsupervised learning methods and event detection techniques for energy disaggregation. After detecting load events, online learning methods will be employed to continuously update the classification model as new appliance classes are identified.

This project will contribute to the development of low-cost, intelligent energy monitoring solutions that can improve energy efficiency, reduce unnecessary power consumption, and enable smarter automation within residential and industrial environments.

Research Context & Motivation

Nowadays, there is a growing trend toward the use of distributed sensors to perform localized measurements, particularly in the context of Industry 4.0. One major issue with this approach, however, is that deploying large numbers of distributed sensors in a factory or building can become highly expensive. In addition, storing, processing, and analyzing such large volumes of data introduce significant computational and financial costs.

To address this, we propose the use of a Non-Intrusive Load Monitoring (NILM) approach, where measurements taken from a single point are used to infer information about an entire distributed electrical system. NILM aims to disaggregate total power consumption into individual appliance-level activities using machine learning and signal analysis techniques.

Here, we propose applying this approach within a dedicated smart house environment. The smart house installation already contains a distributed measurement infrastructure, making it an ideal environment for testing and validating our proposed methods. Beyond smart homes, the developed solutions could also be applied to industrial facilities, commercial buildings, and smart grid systems where there is a need to monitor energy usage without installing large numbers of sensors. By reducing the requirement for distributed hardware, NILM approaches lower installation and maintenance costs while providing detailed information supporting predictive maintenance, fault detection, and intelligent energy management strategies.

Application & Course Details

Interested candidates should review the formal course details below for their application:

Detail Specification
Course Title Computer Science and Informatics (PhD)
Mode of Attendance (MoA) Full-Time (FT)
Admissions Email gs.pgradmissions@hud.ac.uk

Supervisory Team