研究目的
To predict the excitation wavelength of phosphors using machine learning models to reduce the cost and time of experimental approaches in solid-state lighting applications.
研究成果
The machine learning models (LASSO and ANN) successfully predicted excitation wavelengths with high accuracy (R2 > 0.99), demonstrating a cost-effective alternative to experimental methods. Future work should include more features and samples to improve accuracy and address conflicting results.
研究不足
The model is based on a small dataset (49 samples), which may limit generalizability. Some materials showed conflicting results, indicating potential inaccuracies. The approach is restricted to simple molecular structures and may not account for all factors influencing excitation.
1：Experimental Design and Method Selection:
The study uses machine learning models (LASSO and ANN) to predict excitation wavelengths based on atomic-level features of phosphors. The rationale is to leverage easily accessible attributes to avoid lengthy lab experiments.
2：Sample Selection and Data Sources:
Data from 49 luminescent materials with known excitation wavelengths were used, sourced from various references.
3：List of Experimental Equipment and Materials:
No specific equipment or materials are listed for physical experiments, as the study is computational.
4：Experimental Procedures and Operational Workflow:
Feature vectors were created using 15 atomic attributes per atom (host and activator), standardized using Python's StandardScaler, and fed into LASSO and ANN models implemented with sklearn and keras libraries.
5：Data Analysis Methods:
Statistical analysis included mean absolute error, mean squared error, root mean square error, and coefficient of determination (R2) to evaluate model performance.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
