The chiller fault data are often difficult to obtain in the field

which is one of the key obstacles hindering the field applications of chiller fault detection. Considering this reality

the fault detection task is transformed into a typical one-class classification problem by merging a distance rejection (DR) technique into a Bayesian network (BN); therefore

a method based on DR-BN is proposed in this study. The proposed method effectively overcomes the above-mentioned limitation by using the normal data alone to train the model. The performance of the proposed method is evaluated by using the experimental data from the ASHRAE RP-1043

and compared with the other traditional methods. The proposed method shows a better performance than the other traditional methods. Especially for the low serious level

the maximum accuracy of the proposed method is increased by 94%.