This study proposes methods of the multi-split type air-conditioning system model based on refrigerants thermodynamic properties that improve the computational speed reserving accuracy and extrapolability, and reduce physical parameters setting work. The purpose is to improve the quantitative performance assessment of the multi-split type air-conditioning system. The approximation method of refrigerative physical properties value using the neural network improves the computational speed. When the heat load of indoor equipment is input to the multi-split type air-conditioning system model, the amount of consumption energy is the output. It is found the proposed model can calculate approximate outputs in cases of not only JIS but other conditions.
ASJC Scopus subject areas