Synergizing neural networks with multi-objective thermal exchange optimization and PROMETHEE decision-making to improve PCM-based photovoltaic thermal systems
This study addresses the integration of machine learning (ML) and artificial intelligence (AI) for optimizing photovoltaic-thermal (PVT) systems. While ML modeling has become prevalent in this field, a significant gap remains in combining ML with AI-based optimization and decision-making methods to...
Saved in:
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-04-01
|
| Series: | Case Studies in Thermal Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2214157X2500111X |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This study addresses the integration of machine learning (ML) and artificial intelligence (AI) for optimizing photovoltaic-thermal (PVT) systems. While ML modeling has become prevalent in this field, a significant gap remains in combining ML with AI-based optimization and decision-making methods to enhance PVT performance. This research introduces a four-step hybrid framework that integrates data analysis, ML modeling, multi-objective optimization (MOO), and multi-criteria decision-making (MCDM) to achieve comprehensive PVT system optimization. In a case study of a phase change material (PCM)-based PVT system, a GMDH-type ANN model was applied to predict electrical power (EP), thermal power (TP), and entropy generation (EG) based on inputs including PCM melting temperature, PCM thickness, solar radiation, and ambient temperature. Results demonstrated the model's high accuracy, with R2 values exceeding 0.998 across objectives. MOO based on GMDH-NN models using the novel multi-objective thermal exchange optimization (MOTEO) algorithm revealed trade-offs: maximizing EP and TP increased EG, highlighting inherent inefficiencies. The Pareto optimal points highlighted the significant influence of environmental conditions and PCM characteristics, with an ambient temperature of 10 °C, a PCM melting temperature of 25 °C, and a PCM thickness of 1.5 cm yielding the majority of optimal outputs. MCDM through PROMETHEE provided design flexibility, allowing weighted objectives to support various design scenarios. This framework underscores ML and AI's potential to elevate PVT system performance, establishing a foundation for future renewable energy technologies. |
|---|---|
| ISSN: | 2214-157X |