This paper presents a data-mining-based method with a decision tree algorithm to diagnose accumulator opposite-insertion faults for a variable refrigerant flow (VRF) system. First

the VRF experimental platform was established for data collection. Then

expert knowledge and verification methods for the data variation model were used to select appropriate model variables. The C5.0 decision tree algorithm was employed to develop a fault diagnosis model. Finally

the effect of the electronic expansion valve (EEV) of the subcooler was analyzed and validated; this effect was selected as the best variable on the basis of classification rules generated by the model. The results show that the fault diagnosis method based on the decision tree exhibits desirable effectiveness for diagnosing accumulator opposite-insertion faults

with which the fault diagnosis accuracy is up to 96%. Moreover

the proposed technique can meet the requirements for online application of fault diagnosis for VRF systems. This method incorporating a decision tree algorithm to diagnose accumulator opposite-insertion faults for a VRF system exhibits very high accuracy and reliability; therefore

the method can meet the actual demands of fault diagnosis for VRF systems. Because the occurrence of accumulator opposite-insertion faults corresponds to a reduction in the degree of superheating

increasing the opening of subcooler EXV can ensure the cooling effect and a good energy efficiency ratio.