The modeling methods research and comparison of a heat exchanger using neural network

In order to accurately obtain dynamic characteristics of a heat exchanger, neural network technology was used to model it, and obtain black-box model and gray-box model. Based on Back Propagation (BP) algorithm, the two models were respectively trained with real operating data of the heat exchanger. The comparison between the outputs of the two well-trained models and the real output of the heat exchanger shows that the gray-box model is more complicated than the black-box model, but it has less training time and more accurate than the black one.