| Title |
Validation of PCA Method for Failure Detection of Repeater of Measuring Sensor in Chiller |
| Authors |
Kwak, Ro-Yeul ; Seo, Dae-Woong |
| DOI |
http://dx.doi.org/10.5659/JAIK_PD.2015.31.3.181 |
| Keywords |
chiller ; sensor ; fault detection ; PCA ; Q-statistics |
| Abstract |
Building energy management system(BEMS) in order to save the energy used in the operation stage of buildings has been actively introduced recently and it is recommended to apply BEMS to various certification system, but, there is a real situation such as low utilization level of comparison of energy savings and monitoring function oriented simple statistical processing. In this study, by using the principal component analysis(PCA) method that is able to analyze whether failure in repeater of measuring sensor for the chiller measurement data in the actual building where BEMS is operating, failures that exist in BEMS data were detected. PCA method is the most widely used method of adjusting the data and could detect a fault and improve the performance of diagnosis of low-dimensional description of data generated by the PCA method using multivariate statistics such as the Hotelling T2 statistics and Q-statistics to monitor the industrial system. It is found that because the end of operation time before replacing the sensor and Q-statistics beyond Q-Thresholds has been appeared to the same sample position, the failure of the sensor can be detected by PCA method using the stored BEMS data. |