Parameter selection in data-driven fault detection and diagnosis of the air conditioning system

Data-driven fault detection and diagnosis system (FDD) has been proven as simple yet powerful to identify soft abrupt faults in the air conditioning system, leading energy saving. However, challenge is obtain reliable operation data from actual building. Therefore, a lab-scaled centralized chilled water was successfully developed this paper. All necessary sensors were installed generate for data-driven FDD. Nevertheless, if practical considered, number of required would be extensive it depends on rooms Hence, parameters impact dataset also investigated critical classifications. The analysis results had identified four FDD: rooms' temperature (TTCx), supplied (TCHWS), flow rate (VCHWS) cooled (TCWS). Results showed that FDD diagnosed all six conditions correctly with proposed more than 92.3% accuracy; only 0.6-3.4% differed original dataset's accuracy. can reduce used buildings, thus reducing installation costs without compromising