TY - JOUR
T1 - A cosine-based correlation information entropy approach for building automatic fault detection baseline construction
AU - Huang, Jiajing
AU - Yoon, Hyunsoo
AU - Pradhan, Ojas
AU - Wu, Teresa
AU - Wen, Jin
AU - O’neill, Zheng
AU - Candan, Kasim Selcuk
N1 - Publisher Copyright:
© Copyright © 2022 ASHRAE.
PY - 2022
Y1 - 2022
N2 - Building automatic fault detection and diagnosis (AFDD) technologies have shown great potential for energy savings. To enable AFDD, a baseline depicting the normal operation mode is needed to detect whether the building operation deviates from normality. Existing research using physics-based knowledge and models for AFDD has mainly taken a trial-and-error approach to determine if a given baseline is sufficient via empirical experiments. A mechanism to support decisions such as how many samples and what samples should be included in the baseline is currently lacking. In this study, a data-driven method for AFDD baseline construction based on information entropy is developed. The entropy is derived based on cosine similarity among typical building automation system measurements in conjunction with outdoor weather information. The performance of the proposed method is evaluated using real building data. Evaluation results indicate that the fault detection strategy adopting the proposed method has similar or better accuracy in detecting faults compared to the same fault detection strategy using the baseline construction method from the literature. In addition, the use of entropy enables the proposed method to automatically construct and assess the baseline consisting of information-rich samples.
AB - Building automatic fault detection and diagnosis (AFDD) technologies have shown great potential for energy savings. To enable AFDD, a baseline depicting the normal operation mode is needed to detect whether the building operation deviates from normality. Existing research using physics-based knowledge and models for AFDD has mainly taken a trial-and-error approach to determine if a given baseline is sufficient via empirical experiments. A mechanism to support decisions such as how many samples and what samples should be included in the baseline is currently lacking. In this study, a data-driven method for AFDD baseline construction based on information entropy is developed. The entropy is derived based on cosine similarity among typical building automation system measurements in conjunction with outdoor weather information. The performance of the proposed method is evaluated using real building data. Evaluation results indicate that the fault detection strategy adopting the proposed method has similar or better accuracy in detecting faults compared to the same fault detection strategy using the baseline construction method from the literature. In addition, the use of entropy enables the proposed method to automatically construct and assess the baseline consisting of information-rich samples.
UR - http://www.scopus.com/inward/record.url?scp=85131856454&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85131856454&partnerID=8YFLogxK
U2 - 10.1080/23744731.2022.2080110
DO - 10.1080/23744731.2022.2080110
M3 - Article
AN - SCOPUS:85131856454
SN - 2374-4731
VL - 28
SP - 1138
EP - 1149
JO - Science and Technology for the Built Environment
JF - Science and Technology for the Built Environment
IS - 9
ER -