Ongoing energy fault detection using a data-driven chiller performance prediction model

Ongoing energy fault detection is a process of continuously comparing the actual performance of the building system calculated from the current monitoring data with the pre-determined target performance predicted by a mathematical model. In this paper, a noble ongoing energy fault detection method using multiple locally weighted linear regression models is proposed to provide more accurate prediction and reduce false alarms. In order to demonstrate the efficiency of the proposed method, its performance is empirically evaluated over the monitoring data acquired from a real-world centrifugal chiller and compared with the one of previous method in terms of both prediction and detection accuracy.