Author(s): Grant Mosey & Brian Deal
This paper explores the use of new tools for the creation of novel methods of identifying faults in building energy performance remotely. With the rise in availability of interval utility data and the proliferation of machine learning processes, new methods are arising which promise to bridge the gap between architects, engineers, auditors, operators, and utility personnel. Utility use information, viewed with sufficient granularity, can offer a sort of 鈥済enome,鈥漷hat is a set of 鈥済enes鈥 which are unique to a given building and can be decoded to provide information about the building鈥檚 performance. The applications of algorithms to a large data set of these 鈥済enomes鈥 can identify patterns across many buildings, providing the opportunity for identifying mechanical faults in a much larger sample of buildings that could previously be evaluated using traditional methods.
Volume Editors
John Folan & Julie Ju-Youn Kim
ISBN
978-1-944214-13-5