Development of Neural Networks for Enhancement of Thermal Energy Storage using Phase Change Material

Abstract

Three Artificial Neural Network models (Feedforward, Elman, and Nonlinear Autoregressive Exogenous (NARX) networks) were used to find the performance of a thermal energy storage system with and without a phase change material. Previously obtained experimental data was used to train the neural network. Time, mass of water, mass flow rate, number of balls containing the PCM, hourly solar radiation, ambient temperature and inlet water temperature were used in the input layer of the network. The outlet water temperature was in the output layer. The obtained results were verified against previously obtained experimental data. It was found that Artificial Neural Network technique could be used to estimate the outlet temperature with excellent accuracy with the coefficient of determination of Elman, feedforward and NARX models were found to be 0.95006, 0.99411 and 0.88185, respectively. The obtained results showed that feedforward model had the best ability to estimate the required performance, while NARX and Elman network had the lowest ability to estimate it.