Robust Efficient Estimation of Energy Use

Predicting hourly building energy use from environmental input variables can help to increase energy efficiency of buildings, and so contribute to global sustainability. Small Energy Group, a Vancouver-based energy management software company has been working on this challenging prediction problem. Small Energy Group has collected energy data on particular buildings so that the building owners and managers can identify ways to lower their energy consumption. The goal of this MITACS ACCELERATE internship is to provide managers with a “typical energy use curve” based on building type and meteorological input. A typical curve cannot be described by standard statistical methods such as linear regression. Therefore, the internship will develop flexible methods based on a technique called smoothing. Methods must take into account unusual values of energy use, identify which unusual values arise from a malfunction in measurement and which arise because of unusually high energy use.This project is in partnership with the National Institute for Complex Data Structures.

Faculty Supervisor:

Dr. Nancy Heckman

Student:

Camila Pedroso Estevam de Souza

Partner:

Small Energy Group

Discipline:

Statistics / Actuarial sciences

Sector:

Energy

University:

University of British Columbia