Automated Anomaly Detection in Energy Consumption Data using Data Mining and Machine Learning Techniques

The Ontario government recently mandated electric and natural gas utilities provide easy access to energy usage data so consumers can access energy-saving services in a competitive market. However, there has been no analysis of how well utilities have implemented Green Button. More specifically, are regulatory requirements being met, and is data accurate, complete, consistent with other available sources? This lack of scrutiny imperils Ontario’s efforts to decarbonize its economy because poor-quality data can inhibit clean energy solutions such as energy management, demand response, and virtual power plants. This project is the first to evaluate Ontario’s energy usage, billing, and account data quantitatively/qualitatively from utilities to benchmark performance. This project seeks to minimize resource use and time by employing a mixed-method machine learning approach, i.e., semi-supervised learning. The findings from this study will apply to other Canadian provinces.

Faculty Supervisor:

Francis Palma

Student:

Partner:

Screaming Power

Discipline:

Computer science

Sector:

Professional, scientific and technical services

University:

University of New Brunswick

Program: