Abstract
The current article provides a summary of automated fault detection and diagnostics studies published since 2004 that are relevant to the commercial buildings sector. The review updates a previous review conducted in 2004 and published in 2005, and it categorizes automated fault detection and diagnostics methods into three groups. The examples of automated fault detection and diagnostics in the primary category are selectively reviewed to identify various methods that are suitable for building systems and to understand the strengths and weaknesses of the methods. The distribution of studies based on each automated fault detection and diagnostics method and heating, ventilation, and air-conditioning system is also described. Researchers and industries can use the current article as a guideline for selecting an appropriate automated fault detection and diagnostics method.
Nomenclature | ||
AFFD | = | = automated fault detection and diagnostics |
AHU | = | = air-handling unit |
ANN | = | = artificial neural network |
AR | = | = autoregressive |
ARMA | = | = autoregressive-moving–average |
BPNN | = | = back-propagation neural network |
CUSUM | = | = cumulative sum |
GRNN | = | = general regression neural network |
HVAC | = | = heating, ventilation, and air conditioning |
HVAC&R | = | = HVAC and refrigeration |
LED | = | = light-emitting diode |
JAA | = | = joint angle analysis |
PCA | = | = principal component analysis |
PID | = | = proportional integral and derivative |
RNN | = | = recurrent neural network |
RTUs | = | = rooftop units |
SPC | = | = statistical process control |
SAX | = | = symbolic aggregate approximation |
SEER | = | = seasonal energy efficiency ratio |
SVM | = | = support vector machine |
VAV | = | = variable-air volume |
Acknowledgments
The authors thank Dr. Marina Sofos, Technology Development Manager, for her guidance and strong support of this work. At PNNL, the authors acknowledge Susan Ennor for her editorial support in preparing this article.
Funding
The authors acknowledge the Buildings Technologies Office of the U.S. Department of Energy's Office of Energy Efficiency and Renewable Energy (under Contract DE-AC05-76RL01830) for supporting this research and development effort.
Notes
1 Q-statistics is a test statistic output used in statistical hypothesis testing that provides better small-sample properties.