A fault detection and diagnosis strategy with enhanced sensitivity for centrifugal chillers

This paper presents a fault diagnosis strategy based on a simple regression model and a set of generic rules for centrifugal chillers. Four characteristic quantities obtained from low-cost measurements are used as fault indexes. Residuals, which are the differences between the fault indexes calculated from the model and the actual measurements respectively, are used to detect fault. Adaptive thresholds are adopted considering measurement and modeling uncertainties. Relative sensitivity of each fault index to a fault is analyzed so that the most sensitive one is selected and used in the rule-based fault diagnosis. Compared with previous study, the sensitivity of the fault diagnosis method is evidently enhanced by relating each fault to both the direction and the magnitude that the most sensitive index changes when the fault occurs. Seven common faults in typical centrifugal chillers are considered. The strategy is validated using the data sets of ASHRAE research project RP-1043. The results show that the FDD strategy developed is reliable and efficient with lower computation load and higher sensitivity. Therefore, it is quite suitable for online fault diagnosis of centrifugal chillers.