A Self Learning Algorithm for Temperature Prediction in a Single Family Residence

Published

September 29, 2015

Author(s)

Farhad Omar, Steven T. Bushby

Abstract

In order to develop effective control optimization strategies to manage residential electricity consumption in a smart grid environment, predictive algorithms are needed that are simple to implement, minimize custom configuration, and provide enough accuracy to enable meaningful control decisions. Two of the largest electrical loads in a typical residence are heating and air conditioning. A self learning algorithm for predicting indoor temperature changes is derived using a first-order lumped capacitance technique. The algorithm is formulated in such a way that key design details such as window size and configuration, thermal insulation, and air tightness that effect heat loss and solar heat gain are combined into effective parameters that can be learned from observation. This eliminates the need for custom configuration for each residence. Using experimental data from the National Institute of Standards and Technology (NIST) Net Zero Energy Residential Test Facility (NZERTF), it was demonstrated that an effective overall heat transfer coefficient and thermal time constant for the house can be learned from a single nighttime temperature decay test. It was also demonstrated that an effective solar heat gain coefficient can be learned without knowledge of the window area and orientation by application of a self-learning, sliding-window algorithm that accounts for seasonal variations and daily weather fluctuations. The resulting algorithm is shown to be able to predict indoor temperatures for a one-day time horizon using a solar irradiance and outdoor temperature forecast, and control decisions for operating a heat pump.

Citation

Technical Note (NIST TN) - 1891

Report Number

1891

NIST Pub Series

Technical Note (NIST TN)

Pub Type

NIST Pubs

Download Paper

DOI Link

Keywords

Net zero energy house, net zero energy residential test facility, thermal model, lumped capacitance model, thermal time constant, overall heat transfer coefficient, moving window optimization, sliding window optimization, parameter learning, parameter fitting, parameter optimization

Citation

Omar, F. and Bushby, S. (2015), A Self Learning Algorithm for Temperature Prediction in a Single Family Residence, Technical Note (NIST TN), National Institute of Standards and Technology, Gaithersburg, MD, [online], https://doi.org/10.6028/NIST.TN.1891 (Accessed April 19, 2024)