Online sensor fault diagnosis for robust chiller sequencing control

Abstract

Chiller chilled water flow rate, supply and return temperature are used in building cooling load direct measurement in central chilling systems. Healthy sensor measurements of them are essential for proper chiller sequencing control. Site experience indicates that these measurements are easily corrupted by systematic errors or measurement faults. Therefore, an online sensor fault detection and diagnosis (FDD) strategy based on data fusion technology is developed to detect faults in the building cooling load direct measurement. The confidence degree, generated by a data fusion algorithm, is used to indicate the existence of the faults. The faults in the chilled water flow rate and supply temperature measurements are diagnosed according to the redundant information provided in building automation system (BAS). The faults in the return water temperature measurements are diagnosed by reconstructing the confidence degree using the expected values of the chilled water flow rate and the supply temperature by taking account of the associated uncertainties. Cases studies are performed on a simulated central chilling plant equipped in a high-rising building in Hong Kong. The results demonstrate satisfactory effectiveness of the proposed method in diagnosing faults in the building cooling load direct measurement.