Whole Building System Fault Detection Based on Weather Pattern Matching and PCA Method

Multivariate statistical process analysis (MSPA) methods have been widely employed for component level fault detection in buildings. An MSPA method named as weather pattern matching (PM) and principal component analysis (PCA) method is proposed for whole building system fault detection. This method is modified from a component level fault detection method which is proved effective in detecting faults in air handling unit (AHU) and variable air volume (VAV) terminal. In the proposed method, Symbolic Aggregate approXimation (SAX) method is employed to find similar weather pattern in historical database to accurately generate dynamic baseline dataset for PCA model to detect system faults. One real building data is used to evaluate the effectiveness of the proposed method.