A parametric, control-integrated and machine learning-enhanced modeling method of demand-side HVAC systems in industrial buildings: A practical validation study

Dezhou Kong, Yu Hong, Yimin Yang, Tingyue Gu, Yude Fu, Yihang Ye, Weihao Xi and Zhiang Zhang

Applied Energy, 2025, vol. 379, issue C, No S0306261924023559

Abstract: The development of high-tech manufacturing in recent years has promoted the rapid growth of industrial cleanrooms. The strict requirements for indoor environment control in cleanrooms lead to huge cooling energy requirements, which has given rise to research on energy modeling of cooling systems relevant to the industrial sector. Efficient and accurate modeling of demand-side Heating, Ventilation and Air-conditioning (HVAC) systems can significantly enhance water-side and air-side control strategies, thereby improving overall energy efficiency. However, existing grey-box methods usually only consider independent air-side or room models in industrial scenarios, and the inefficient handling of missing facility data makes it difficult to achieve optimal model performance. Meanwhile, there is also a lack of systematic investigation on the performance of black-box and grey-box methods in industrial demand-side HVAC modeling scenarios. In this study, based on a real industrial park, a demand-side HVAC system model was established using both the Modelica-based method and pure data-driven method to predict the cooling load under summer conditions. By using a machine learning approach, the true value of the missing internal heat gain is inferred instead of using a fixed value. The model results after Genetic Algorithm (GA) calibration show that in the case of small sample size, the prediction error of the machine learning-enhanced Modelica-based model is reduced by about 34.5 % compared with the original modelica-based model. Compared with the data-driven model with hyperparameter tuning, the reduction is about 38.8 %/21.3 % in the case of small/large sample size, respectively. Then the reasons for the differences in the performance of these modeling methods are compared and discussed in detail. The framework of machine learning-enhanced modeling method proposed in this study can also be applied to other industrial demand-side HVAC system modeling scenarios with missing on-site facility data, ultimately achieving more efficient industrial production through optimal control strategies for water-side systems.

Keywords: Industrial buildings; Modelica; Sensitivity analysis; Machine learning; Parametric modeling (search for similar items in EconPapers)
Date: 2025
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0306261924023559
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:appene:v:379:y:2025:i:c:s0306261924023559

Ordering information: This journal article can be ordered from
http://www.elsevier.com/wps/find/journaldescription.cws_home/405891/bibliographic
http://www.elsevier. ... 405891/bibliographic

DOI: 10.1016/j.apenergy.2024.124971

Access Statistics for this article

Applied Energy is currently edited by J. Yan

More articles in Applied Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().