Conference Paper

Proceedings of Building Simulation 2019: 16th Conference of IBPSA

High Resolution Model-based HVAC Fault Detection and Diagnosis (FDD) Considering Building Operation Uncertainty

Yifu Shi, Godfried Augenbroe
Georgia Institute of Technology, United States of America

DOI: https://doi.org/10.26868/25222708.2019.210453
Abstract: Along with the development of data analysis, the datadriven fault detection and diagnoses (FDD) method becomes very popular. However, the data-driven FDD methods requires a large amount of data from Building Automation System (BAS), and these data are usually unreliable. In this study, a co-simulated model which integrated the building energy model with HVAC detail model is developed in the co-simulated platform – Building Controls Virtual Test Bed (BCVTB). The white box co-simulation model can provide reliable trending data to develop and validate the data-driven FDD algorithm. Since the deterministic model is not sufficient to reflect the actual operation conditions, the building physical and operational uncertainty is considered in this study by using Monte Carlo method. Typical AHU faults are configured in the model, and an FDD algorithm is developed and validated based on the specific building operation condition.
Keywords: Modelica, Co-simulation, FDD, Uncertainty, HVAC
Pages: 1832 - 1840
Paper: