4589 -- The Use of Time Series Analysis in Fault Detection and Diagnosis Methodologies PDF

Existing and proposed fault detection and diagnosis methodologies for HVAC systems and components contain statistical and thermodynamic analysis techniques. However, few have evaluated the use of time series analysis to detect faults. An in depth examination of a data set collected from a commercial chiller plant was analyzed. The detailed analysis revealed unexpected time varying trends in the data. Changes in the trends indicated the existence of “faults,” which are unexplained changes in the system operating characteristics. The use of time series analysis and the potential benefits of including such techniques in chiller fault detection and diagnosis are discussed.

Units: I-P