Using Pattern Matching and Principal Component Analysis Method for Whole Building Fault Detection (LB-17-C044)

Automated fault detection and diagnosis (AFDD) methods, followed by corrections, have the potential to greatly improve a building and its system’s performances. Existing AFDD studies mostly focus on component and sub-system AFDD. Much less effort has been spent on detecting and diagnosing faults that have a whole building impact. In this paper, an integrated data driven method: Pattern Matching Principle Component Analysis method, is developed and applied for whole building fault detection. Real building data that contains artificially injected faults and naturally occurred faults are used to evaluate the method’s accuracy and false alarm rate. The method presents great potential to be a cost-effective and accurate whole building fault detection strategy.