ESTIMATING THE SIZE OF INCIPIENT FAULTS IN AIR-CONDITIONING SYSTEMS

An approach to identifying incipient faults in air-conditioning systems is proposed, which is based on a fuzzy sliding mode observer that estimates the system output and forms an estimate of the fault using an equivalent control signal. A neurofuzzy technique is used to build a fuzzy relational model of the non-linear dynamic behaviour of the system. The main advantage of using a sliding mode observer is that the output estimator is robust to modeling errors and disturbances. The equivalent control signal is composed of two parts: one is the control signal required to track the system output; the other part compensates for the fault. The proposed fuzzy identification method is evaluated on the cooling coil subsystem of an air-conditioning plant. Simulation results are presented which demonstrate its effectiveness.