Scholarlink

<p>The evolution in photovoltaic (PV) technologies has seen significant advancement with the development of third-generation solar cells, i.e., organic photovoltaics (OPVs), perovskite solar cells (PSCs), and dye-sensitized solar cells (DSSCs). These technologies promise enhanced efficiency, flexibility, and cost-reduction, positioning them capable in the worldwide tendency towards sustainable energy production. Italy has become a key player in development PV technology, and also in developing high degree innovation, with regional research centres significantly involved in new material development, building and indoor PV optimization, and integration in the already existing renewable energy sources. We listed the thermophysical properties of materials used in PV cells for various functions: ensuring uniform temperature of solar radiation materials through thermal conductivity; controlling heat transfer within the cell; and measuring heat absorption and release <i>via</i> specific heat. Consequently, methods for measuring these properties in thin layers and coatings are essential for characterizing cell behaviour. In technical aspect, this paper presents machine learning methods to predict power-conversion efficiency (PCE) and thermal conductivity in a small and large dataset of OPV, PSC and DSSC solar cell materials. It emphasizes the use of artificial neural network (ANN), K-nearest neighbours (KNN), support vector machine (SVM), and decision tree algorithms on PSC, DSSC, and OPV. To evaluate the algorithms performance, we us two datasets of varying size on prediction of PCE with regression analysis. Dataset ‘A’ with 6 features and 1000 samples suggesting that ANN and SVM models performed better with a root-mean-square error (RMSE) of around 0.54 and coefficient of determination (R<sup>2</sup>) of 0.75. Dataset ‘B’ with 10 000 samples and same features exhibited improvement in the model generalization; the ANN showed good predictive capability, with RMSE = 0.698 and R<sup>2</sup> = 0.956. The results of these algorithms highlight the role of optimal design of PCE through the incorporation of knowledge of thermal conductivity along with other electrical and optical material properties during solar cell design. Moreover, this work outlines the synergies between photovoltaic innovations emerging in different regions of Italy, and the advantages obtainable from them, such as stability, long-term use, cost-reduction, scalability and increased efficiency. So it underlines how these new technologies can accelerate their diffusion through energy communities, by means of low-cost, high-efficiency, and application-flexibility of future solar systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Numerical and Machine Learning Methods to Predict the Thermal Behaviour of New Photovoltaic Cells in Order to Increase Their Efficiency

  • 期刊论文

  • 发表日期：

    2026年3月21日

查看全文
  • Ateeq Ur Rehman,
  • Sandra Corasaniti,
  • Ivano Petracci,
  • Paolo Coppa,
  • Dario Atzori

摘要

The evolution in photovoltaic (PV) technologies has seen significant advancement with the development of third-generation solar cells, i.e., organic photovoltaics (OPVs), perovskite solar cells (PSCs), and dye-sensitized solar cells (DSSCs). These technologies promise enhanced efficiency, flexibility, and cost-reduction, positioning them capable in the worldwide tendency towards sustainable energy production. Italy has become a key player in development PV technology, and also in developing high degree innovation, with regional research centres significantly involved in new material development, building and indoor PV optimization, and integration in the already existing renewable energy sources. We listed the thermophysical properties of materials used in PV cells for various functions: ensuring uniform temperature of solar radiation materials through thermal conductivity; controlling heat transfer within the cell; and measuring heat absorption and release via specific heat. Consequently, methods for measuring these properties in thin layers and coatings are essential for characterizing cell behaviour. In technical aspect, this paper presents machine learning methods to predict power-conversion efficiency (PCE) and thermal conductivity in a small and large dataset of OPV, PSC and DSSC solar cell materials. It emphasizes the use of artificial neural network (ANN), K-nearest neighbours (KNN), support vector machine (SVM), and decision tree algorithms on PSC, DSSC, and OPV. To evaluate the algorithms performance, we us two datasets of varying size on prediction of PCE with regression analysis. Dataset ‘A’ with 6 features and 1000 samples suggesting that ANN and SVM models performed better with a root-mean-square error (RMSE) of around 0.54 and coefficient of determination (R2) of 0.75. Dataset ‘B’ with 10 000 samples and same features exhibited improvement in the model generalization; the ANN showed good predictive capability, with RMSE = 0.698 and R2 = 0.956. The results of these algorithms highlight the role of optimal design of PCE through the incorporation of knowledge of thermal conductivity along with other electrical and optical material properties during solar cell design. Moreover, this work outlines the synergies between photovoltaic innovations emerging in different regions of Italy, and the advantages obtainable from them, such as stability, long-term use, cost-reduction, scalability and increased efficiency. So it underlines how these new technologies can accelerate their diffusion through energy communities, by means of low-cost, high-efficiency, and application-flexibility of future solar systems.