Dynamic modeling of absorption chillers using LSTM and CNN-LSTM.
DOI:
https://doi.org/10.17979/ja-cea.2026.47.13696Keywords:
LSTM Model, CNN-LSTM Hybrid Model, Absorption MachineAbstract
The following research presents the development of “black box” models for the dynamic characterization of a Yazaki WFC SC20 absorption chiller. Unlike previous studies based on proprietary software, this research uses open-source software technologies such as Python and PyTorch to improve the reproducibility and scalability of the modeling. Two deep neural network architectures were designed and compared: a Long Short-Term Memory (LSTM) network and a hybrid convolutional-recurrent model (CNN-LSTM). To optimize performance, the Optuna framework was used, enabling intelligent hyperparameter search and input variable selection. The models were trained and validated with real data from the CIESOL center’s solar thermal installation from 2014 and 2015. The results demonstrate that the architecture that best describes the absorption chiller is the CNN-LSTM.
References
Akiba, T., Sano, S., Yanase, T., Ohta, T., Koyama, M., 2019. Optuna: A next generation hyperparameter optimization framework. In: Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. KDD ’ 19. Associationfor Computing Machinery, NewYork, NY, USA, p. 2623–2631. URL: https://doi.org/10.1145/3292500.3330701 DOI: 10.1145/3292500.3330701
Ferre, A., Castilla, M., Carballo, J. A., ´Álvarez, J. D., 2023. Characterization of an absorption machine using artificial neural networks. In: Maglogiannis, I., Iliadis, L., MacIntyre, J., Dominguez, M. (Eds.), Artificial Intelligence Applications and Innovations. Springer Nature Switzerland, Cham, pp. 279–291.
Gobierno de España, 2023. Plan nacional integrado de energía y clima 2023 2030 (pniec). Documento oficial.
Hosseini, P., 2022. Deep-learning neural network prediction of a solar-based absorption chiller cooling system performance using waste heat. Sustainable Energy Technologies and Assessments 53, 102683. URL: https://www.sciencedirect.com/science/article/pii/S2213138822007329 DOI: https://doi.org/10.1016/j.seta.2022.102683
IEA, 2022. International energy agency- key world energy statistics 2021. Tech. rep. URL: https://www.iea.org/reports/key-world-energy-statistics-2021
Panahizadeh, F., Hamzehei, M., Farzaneh-Gord, M., Villa, A. A. O., 2021. Evaluation of machine learning-based applications in forecasting the performance of single effect absorption chiller network. Thermal Science and Engineering Progress 26, 101087. URL: https://www.sciencedirect.com/science/article/pii/S2451904921002481 DOI: https://doi.org/10.1016/j.tsep.2021.101087
Rodríguez-Ramos, A., Ma, L., Verde, C., Silva Neto, A. J., Llanes-Santiago, O., 2026. Anoptimized deep learning framework for fault and cyberattack detection in chemical process monitoring. Chemical Engineering Science 320, 122608. URL: https://www.sciencedirect.com/science/article/pii/S0009250925014290 DOI: https://doi.org/10.1016/j.ces.2025.12260
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Zuliet Medina Rodríguez, Juana López Redondo , María del Mar Castilla Nieto, José Domingo Álvarez Hervás
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
