Skip to Main Content
To accurately obtain dynamic characteristics of a heat exchanger, black-box modeling method and gray-box modeling method were used with the help of neural network technology. The black-box model directly used the heat exchanger's input and output data to train the neural network. It constantly adjusted the network's weight to record the system's dynamic characteristics, and then predict output. Having known the heat exchanger's operating point, the gray-box model used filtered input and output data to train the neural network. It also identified the model parameters at the operating point, and then predicted output. By comparing experimental data of the two models, the results show that the gray-box model is better than the black-box model on hidden layer nodes, network training, errors and prediction accuracy.