Sensor Fault Diagnosis Method Based On KPCA In Constant Air-volume Air-conditioning Unit

Sensor fault is one of the typical faults in Air-conditioning system. The fault sensor's signal will make the Air-conditioning system to do a wrong decision-making, thus causing the energy waste and descend of indoor environment quality, so it is meaningful to research the sensor fault diagnosis.The existent methods of Air-conditioning system's sensor fault diagnosis are summarized and the advantages and disadvantages of every method are analyzed. Compare kernel principal component analysis(KPCA) to principal component analys- is (PCA), the superiority of KPCA for nonlinear system is deeply analyzed. Aiming at the nonlinear relationship in process variables in Air-conditioning system, a fault diagnosis method is developed based on KPCA.The energy balance relation and the enthalpy in constant air-volume (CAV) Air-conditioning system are analyzed, on the basis of that, the KPCA model and the whole scheme of sensor fault diagnosis method in constant Air-volume Air-conditioni- ng unit based on KPCA is set up.The fault patterns of Air-conditioning system's sensor are analysed, on that basis, the characters of Air-conditioning system's sensor faults such as bias , drift, complete and precision grade decrease faults are deeply researched. Aiming at the four typical faults and using the operating data of Tianjin museum building control system in normal condition, simulations are respectively performed for KPCA and PCA. Judging from the comparison of simulation results, KPCA has better effect on Air-conditioning system's sensor detection. The sensor fault is identified by KPCA contribution plots and simulation result proved that it has good effect of fault identification.