Explainable artificial intelligence for energy efficiency in experimental refrigeration Systems: Advanced cutting-edge sunflower optimization
This study applies a cutting-edge artificial intelligence model to an experimental refrigeration system operating with R290, R1234yf, R404A, and R134A refrigerants. In the first part of the study, a comparison was made on indicators such as COP, cooling capacity and compressor power consumption. The...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-09-01
|
| Series: | Case Studies in Thermal Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2214157X2500841X |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This study applies a cutting-edge artificial intelligence model to an experimental refrigeration system operating with R290, R1234yf, R404A, and R134A refrigerants. In the first part of the study, a comparison was made on indicators such as COP, cooling capacity and compressor power consumption. The results showed that the highest COP and cooling capacity was provided by R290, and the lowest consumption was achieved with R134A. To create a dataset for artificial intelligence application, evaporator temperatures varying from −17 °C to −3 °C and condenser temperatures varying from 23 °C to 43 °C were used as operating conditions in the experiments. In the second part, experimental data obtained with different refrigerants from the refrigeration system were used to classify compressor power consumption as high, medium, and low. With the proposed rule-based advanced sunflower optimization algorithm (RbA-SFO), the model's high performance and interpretability are intended for comprehensibility and explainability. The RbA-SFO algorithm results were compared with standard rule extraction methods and classification methods. The RbA-SFO achieved superior performance compared to other standard methods by achieving 83 % accuracy for R290 gas, 79.73 % accuracy for R134A gas, 83.54 % accuracy for R1234yf gas, and 87.34 % accuracy for R404A gas. The model used is an explainable artificial intelligence model and has been applied to a refrigeration system data for the first time in the literature. |
|---|---|
| ISSN: | 2214-157X |