An effective predictor of the dynamic operation of latent heat thermal energy storage units based on a non-linear autoregressive network with exogenous inputs: data

This repository contains datasets and source code to develop an artificial intelligence (AI) based tool to estimate the state-of-charge (SoC) and heat transfer fluid (HTF) outlet temperature of a latent heat thermal energy storage (LHTES) tank with water-ice mixture as the phase change material. The AI tool is a non-linear autoregressive network with exogenous inputs (NARX) model, that takes as input the known values of HTF mass flow rate and inlet temperature for the LHTES tank to predict its SoC and HTF outlet temperature. The model recursively uses its past predictions of SoC and HTF outlet temperature to predict the future outputs of these quantities. The dataset comprises CSV files for training and testing the NARX model. In the files for training (Training_dataset.csv) and testing (Test_profile_**.csv , where ** ranges from 9 to 19), the first column represents time in s, the second column represents SoC with no unit and ranging between 0 (representing completely melted ice) and 1 (representing completely solidified ice), the third column represents change in SoC between two time-steps, the fourth column represents HTF outlet temperature in °C, the fifth column represents HTF mass flow rate in kg/s, and the sixth column represents HTF inlet temperature in °C. The source code for the NARX model is provided as a Jupyter Notebook compatible python script (“NARX_final_codes.ipynb”) and a simple text file (“NARX_final_codes.txt”). Detailed explanation of each block of code is provided as comments within the source code itself. The file named “feedback.csv" contains reference values of SoC and HTF outlet temperature, which can be obtained from external sources (eg. sensors or a physics-based model) and compared with the NARX model at certain intervals of time-steps to monitor the model's accuracy over long-term operations. For a more detailed explanation of the dataset and the source code's development and performance, refer to the article "An effective predictor of the dynamic operation of latent heat thermal energy storage units based on a non-linear autoregressive network with exogenous inputs" published in the journal Applied Energy in 2024.

Research results based upon these data are published at https://doi.org/10.1016/j.apenergy.2024.122697