To evaluate an air source heat pump unit’s operation performance in the frosting-defrosting process efficiently and to explore its operational performance under frosting conditions

the loss coefficient of the nominal heating capacity is proposed in this paper. Lab and field tests to date are conducted to establish a loss coefficient for the nominal heating capacity model based on generalized regression neural network. The result of the research shows that the model’s correlation coefficient |r|

is higher than 0.9 and the expected error percentage (EEP) is lower than 6.5%. These results indicate that the model in this paper could be successfully used in predicting operation performance in the frosting-defrosting process of the air-source heat pump. The simulation result can be used in searching for an optimized defrosting control point under different frosting-defrosting conditions.