ANN Methods for COP Prediction of Supermarket Refrigeration System

Abstract

Refrigeration system is one of the main energy consumers. Since COP prediction is the base of optimization analysis of energy saving, ANN models were developed by using on-site testing data to predict COP of the refrigeration system in this paper. BP and RBF neural network were applied to the prediction models respectively, and genetic algorithm was used to optimize the BP network. The prediction results of the models show that the GA-BP model has higher accuracy and better stability than the BP network only. And the RBF model has the smallest error and shortest time on calculation. Both BP and RBF models can apply in COP prediction of the refrigeration systems, but performance of the RBF model is better.