An interpretable and stacking ensemble model for predicting heat and mass transfer of desiccant wheel
The Solid desiccant air conditioning system is popular due to their thorough dehumidification, low energy consumption, and independent temperature and humidity control. In order to design, simulate and optimize the solid desiccant air conditioning system, an accurate model of desiccant wheel must be...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-03-01
|
| Series: | Results in Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2590123025002671 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832582101346549760 |
|---|---|
| author | Mengyang Li Liu Chen |
| author_facet | Mengyang Li Liu Chen |
| author_sort | Mengyang Li |
| collection | DOAJ |
| description | The Solid desiccant air conditioning system is popular due to their thorough dehumidification, low energy consumption, and independent temperature and humidity control. In order to design, simulate and optimize the solid desiccant air conditioning system, an accurate model of desiccant wheel must be established. A prediction model by stacking ensemble learning for desiccant wheel is presented. The model uses an integration approach, Light Gradient Boosting Machine, Random Forest and Back Propagation Neural Network models are used as the first-level base models to learn the data, and the Linear Regression model as a meta-model integrates the output of the base model to obtain the final prediction results. Each base model uses Bayesian optimization for hyperparameter tuning. In addition, the contribution of each input feature and the decision-making process of the model are analyzed using the Shapley additive explanations. To evaluate the model, 13,095 data sets of desiccant wheel operation data were collected. The focus was on predicting the outlet temperature and humidity ratio. Results indicate that the proposed stacking model has notably enhanced its forecasting ability for these parameters. Coefficient of Determination (R2), Root Mean Squared Error (RMSE), and Mean Absolute Error (MAE) were used to measure this stacking model: the process side outlet temperature (R2 = 0.9467, RMSE=1.5239, and MAE = 1.2721), the process side outlet humidity ratio (R2 = 0.9743, RMSE = 0.5728, MAE = 0.4531). In conclusion, the proposed stacking model can provide reliable theoretical guidance for the solid air conditioning system with desiccant wheel as the core. |
| format | Article |
| id | doaj-art-75177c52e3f54f81ae4240e7907f3fc9 |
| institution | Kabale University |
| issn | 2590-1230 |
| language | English |
| publishDate | 2025-03-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Results in Engineering |
| spelling | doaj-art-75177c52e3f54f81ae4240e7907f3fc92025-01-30T05:14:53ZengElsevierResults in Engineering2590-12302025-03-0125104181An interpretable and stacking ensemble model for predicting heat and mass transfer of desiccant wheelMengyang Li0Liu Chen1Energy School, Xi'an University of Science and Technology, Yanta Road Xi'an 710054, ChinaCorresponding author.; Energy School, Xi'an University of Science and Technology, Yanta Road Xi'an 710054, ChinaThe Solid desiccant air conditioning system is popular due to their thorough dehumidification, low energy consumption, and independent temperature and humidity control. In order to design, simulate and optimize the solid desiccant air conditioning system, an accurate model of desiccant wheel must be established. A prediction model by stacking ensemble learning for desiccant wheel is presented. The model uses an integration approach, Light Gradient Boosting Machine, Random Forest and Back Propagation Neural Network models are used as the first-level base models to learn the data, and the Linear Regression model as a meta-model integrates the output of the base model to obtain the final prediction results. Each base model uses Bayesian optimization for hyperparameter tuning. In addition, the contribution of each input feature and the decision-making process of the model are analyzed using the Shapley additive explanations. To evaluate the model, 13,095 data sets of desiccant wheel operation data were collected. The focus was on predicting the outlet temperature and humidity ratio. Results indicate that the proposed stacking model has notably enhanced its forecasting ability for these parameters. Coefficient of Determination (R2), Root Mean Squared Error (RMSE), and Mean Absolute Error (MAE) were used to measure this stacking model: the process side outlet temperature (R2 = 0.9467, RMSE=1.5239, and MAE = 1.2721), the process side outlet humidity ratio (R2 = 0.9743, RMSE = 0.5728, MAE = 0.4531). In conclusion, the proposed stacking model can provide reliable theoretical guidance for the solid air conditioning system with desiccant wheel as the core.http://www.sciencedirect.com/science/article/pii/S2590123025002671Desiccant wheelData-driven modelingStacking ensemble learningPerformance evaluationShapley additive explanations |
| spellingShingle | Mengyang Li Liu Chen An interpretable and stacking ensemble model for predicting heat and mass transfer of desiccant wheel Results in Engineering Desiccant wheel Data-driven modeling Stacking ensemble learning Performance evaluation Shapley additive explanations |
| title | An interpretable and stacking ensemble model for predicting heat and mass transfer of desiccant wheel |
| title_full | An interpretable and stacking ensemble model for predicting heat and mass transfer of desiccant wheel |
| title_fullStr | An interpretable and stacking ensemble model for predicting heat and mass transfer of desiccant wheel |
| title_full_unstemmed | An interpretable and stacking ensemble model for predicting heat and mass transfer of desiccant wheel |
| title_short | An interpretable and stacking ensemble model for predicting heat and mass transfer of desiccant wheel |
| title_sort | interpretable and stacking ensemble model for predicting heat and mass transfer of desiccant wheel |
| topic | Desiccant wheel Data-driven modeling Stacking ensemble learning Performance evaluation Shapley additive explanations |
| url | http://www.sciencedirect.com/science/article/pii/S2590123025002671 |
| work_keys_str_mv | AT mengyangli aninterpretableandstackingensemblemodelforpredictingheatandmasstransferofdesiccantwheel AT liuchen aninterpretableandstackingensemblemodelforpredictingheatandmasstransferofdesiccantwheel AT mengyangli interpretableandstackingensemblemodelforpredictingheatandmasstransferofdesiccantwheel AT liuchen interpretableandstackingensemblemodelforpredictingheatandmasstransferofdesiccantwheel |