The problem of diagnosing faults in air-conditioning systems is of increasi
ng importance. A brief review of heating, ventilating and air-conditioning
(HVAC) fault diagnosis schemes is given, the key design issues are identifi
ed and the advantages of the fuzzy model-based approach are summarised. The
way in which the fuzzy fault diagnosis scheme is able to take account of m
odelling errors, ambiguous data and sensor bias is explained. A multi-step
approach to the fuzzy model-based approach is proposed that makes the resul
ts of the diagnosis less ambiguous. A computer simulation study is used to
demonstrate that a more precise diagnosis can be obtained using the new sch
eme. Experimental results are also presented that show that the proposed sc
heme does not generate false alarms when it is used to commission an air-ha
ndling unit in a commercial office building.