A method for COP prediction of an on-site screw chiller applied in cinema

Abstract

An artificial neural (ANN) model is developed under on-site operating conditions, and it is different within a laboratory. On-site testing carried out to gather amount of data is for the purpose of training and prediction. Back propagation algorithm (BP) is applied to establish the transient coefficient of performance (COP) prediction model. The most suitable network structure is selected by comparison and cross-validation method, and the selected model is trained through setting the suitable network parameters. The error between the predicted results and the experimental data is almost within ±5.0% and the maximum error is 5.8%. Additionally, the validated model is used to evaluate the performance of a water-cooled variable frequency screw water chiller under various operating conditions and different openness of electronic expansion valve (EXV). It is valuable to apply neutral network on water-cooled screw chillers with maximum efficiency when satisfying a building cooling load instantly under necessary conditions.