Scholarlink

<p>The application of machine learning (ML) across diverse domains provides rapid advancements in various technologies by providing a framework for identification and a systematic approach to understanding material properties and their behaviour. ML in photovoltaic (PV) devices introduces a step-by-step methodology by which we can analyse the behaviour of a proposed PV device by comparing it with a comprehensive dataset. This study provides an 11-step methodology for implementing an ML model in simulating a double perovskite Cs<sub>2</sub>Ti(Cl<sub>1−<i>x</i></sub>Br<sub><i>x</i></sub>)<sub>6</sub>-based solar cell, fluorine-doped tin oxide (FTO)/CdS/Cs<sub>2</sub>Ti(Cl<sub>1−<i>x</i></sub>Br<sub><i>x</i></sub>)<sub>6</sub>/CBTS. Using SCAPS-1D, a comprehensive dataset comprising 3300 data points including variations in bandgaps, absorber thickness, shallow donor density, and total defect density was curated. Our results indicate that the ML model exhibits superior prediction reliability for double perovskite Cs<sub>2</sub>Ti(C<sub>1−<i>x</i></sub>Br<sub>x</sub>)<sub>6</sub> device parameters, with a root mean square error (RMSE) of 0.0045, an R-squared value of 0.999, and a cross-validation score of 0.99. The systematic use of various models including support vector regression (SVR), random forest, stacking SVR+RF, and XGBoost provides a deeper understanding of the interdependence of key performance parameters of a PV device.</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

Application of Machine Learning for Performance Analysis and Composition Tuning in Mixed Halide Cs2Ti(Cl1−xBrx)6 Solar Devices

  • 期刊论文

  • 发表日期：

    2025年10月27日

查看全文
  • Vipul Pamwani,
  • Jaspinder Kaur,
  • Rikmantra Basu,
  • Ajay Kumar Sharma,
  • Jaya Madan,
  • Rahul Pandey

摘要

The application of machine learning (ML) across diverse domains provides rapid advancements in various technologies by providing a framework for identification and a systematic approach to understanding material properties and their behaviour. ML in photovoltaic (PV) devices introduces a step-by-step methodology by which we can analyse the behaviour of a proposed PV device by comparing it with a comprehensive dataset. This study provides an 11-step methodology for implementing an ML model in simulating a double perovskite Cs2Ti(Cl1−xBrx)6-based solar cell, fluorine-doped tin oxide (FTO)/CdS/Cs2Ti(Cl1−xBrx)6/CBTS. Using SCAPS-1D, a comprehensive dataset comprising 3300 data points including variations in bandgaps, absorber thickness, shallow donor density, and total defect density was curated. Our results indicate that the ML model exhibits superior prediction reliability for double perovskite Cs2Ti(C1−xBrx)6 device parameters, with a root mean square error (RMSE) of 0.0045, an R-squared value of 0.999, and a cross-validation score of 0.99. The systematic use of various models including support vector regression (SVR), random forest, stacking SVR+RF, and XGBoost provides a deeper understanding of the interdependence of key performance parameters of a PV device.