Most of the existing data-driven fault diagnosis methods are based on the black box model. Although their accuracy is high

it is difficult to explain the diagnosis process and result. In view of the aforementioned problems

this paper proposes a method for fault diagnosis of chillers and the interpretation of diagnosis mechanism based on associative classification. This method

on the premise of high accuracy

performs reverse analysis of diagnostic process based on the rule library in the diagnostic model and explains the mechanism of the faults and process of diagnosis

and thereby increases reliability of the method for diagnosis based on a data-driven mechanism. Experimental data from ASHRAE research project 1043 (ASHRAE rp-1043) was used for model validation of the system. The results indicated that this method

based on associative classification

can effectively identify seven typical chiller faults

and the average accuracy is as high as 90.84%. The important rules extracted in this study coincide well with refrigeration principles and the knowledge of thermodynamic cycles. Thus

the rules can be used for conducting further studies on fault mechanism analysis and fault diagnosis.