Integrated building fault detection and diagnosis using data modeling and Bayesian networks.

Heating, ventilation, and air-conditioning (HVAC) equipment faults and operational errors result in comfort issues and waste of energy in buildings. An Automatic Fault Detection and Diagnosis (AFDD) tool could help facility managers fix comfort and energy issues more efficiently, by identifying the most probable root causes. Existing AFDD methods mostly focus on equipment-level fault detection and diagnostics; almost no attention is given to building level fault diagnosis, considering inter-dependency between equipment through the energy distribution chain. In this work we propose a methodology to automatically derive a Bayesian network from HVAC system topology description such as Haystack. This Bayesian network models and estimates the state of all elements in the system, helping users to identify the most probable root fault. As it is able to ingest evidence from any source (field data, operators, or other models) and is capable of updating its estimates when new evidence is delivered, such a tool could have a great potential to be used interactively on the field. We applied the proposed methodology on simulated and real-world buildings and present in this paper one specific case. • Transfer building system topology and expert knowledge to a Bayesian network. • Transform equipment-level fault detection into building-wide fault diagnosis. • Experiments on simulated and real-world buildings. [ABSTRACT FROM AUTHOR]