PSO based TCN hybrid optimization for turbulent heat transfer prediction of multiple synthetic jets in crossflow
| dc.authorid | 0000-0003-2654-0702 | |
| dc.authorid | 0000-0001-7844-3168 | |
| dc.contributor.author | Akçay, Selma | |
| dc.contributor.author | Buyrukoğlu, Selim | |
| dc.contributor.author | Akdağ, Ünal | |
| dc.contributor.author | Güngör, Bekir | |
| dc.date.accessioned | 2025-09-19T06:51:22Z | |
| dc.date.available | 2025-09-19T06:51:22Z | |
| dc.date.issued | 2025 | |
| dc.department | Mühendislik Fakültesi | |
| dc.description.abstract | The development of artificial intelligence (AI) models offers significant advantages in laborious, expensive, and long-term experimental and numerical studies. Therefore, AI methods have recently attracted considerable attention in different engineering applications. In the present work, the heat transfer performance of multiple synthetic jets immersed in crossflow on a flat target surface was researched experimentally and estimated with different deep learning models (LSTM, BiLSTM, CNN, GRU, and TCN) and particle swarm optimization (PSO). In the experiments, the effects of the main flow Reynolds number (6000 ≤ Re ≤ 40,000), oscillation amplitude (0.22 ≤ Ao ≤ 0.88), and Womersley number (11 ≤ Wo ≤ 27) on the cooling performance were analyzed. The Nusselt number (Nu), friction factor (f), and thermohydraulic performance (THP) were calculated for different flow and jet parameters. The results showed that for Re = 6000, Wo = 27, and Ao = 0.88, the synthetic jet improved the heat transfer by 2.74 times compared with the steady flow. The THP values increase with increasing synthetic jet parameters, and the highest THP is found 2.06 for Re = 6000, Wo = 27 and Ao = 0.88. Among the deep learning models based on experimental data, the TCN algorithm performed the best when optimized with PSO on data processed by COPULA. Compared with the other deep learning models, the hybrid TCN-PSO model achieves lower MAEs for Nu (5.68561), f (0.00725), and THP (0.0356). This research indicates that integrating deep learning with optimization methods is highly effective for designing models of dynamic thermal systems. | |
| dc.identifier.doi | 10.1007/s40997-025-00921-y | |
| dc.identifier.issn | 22286187 | |
| dc.identifier.scopus | 2-s2.0-105015075014 | |
| dc.identifier.scopusquality | Q2 | |
| dc.identifier.uri | https://doi.org/10.1007/s40997-025-00921-y | |
| dc.identifier.uri | https://hdl.handle.net/20.500.12451/14492 | |
| dc.indekslendigikaynak | Scopus | |
| dc.institutionauthor | Akdağ, Ünal | |
| dc.institutionauthor | Güngör, Bekir | |
| dc.institutionauthorid | 0000-0002-1149-7425 | |
| dc.institutionauthorid | 0000-0002-1330-0008 | |
| dc.language.iso | en | |
| dc.publisher | Springer Nature | |
| dc.relation.ispartof | Iranian Journal of Science and Technology - Transactions of Mechanical Engineering | |
| dc.relation.publicationcategory | Makale - Uluslararası Hakemli Dergi - Kurum Öğretim Elemanı | |
| dc.rights | info:eu-repo/semantics/openAccess | |
| dc.subject | Deep Learning Approach | |
| dc.subject | Multiple Synthetic Jets | |
| dc.subject | Particle Swarm Optimization | |
| dc.subject | Turbulent Heat Transfer | |
| dc.title | PSO based TCN hybrid optimization for turbulent heat transfer prediction of multiple synthetic jets in crossflow | |
| dc.type | Article |
