Scholarlink

<p>Accurate energy forecasting is essential for improving the sustainability of the photovoltaic (PV) systems. This study utilizes six years of meteorological data from Duzce to predict solar energy output using 14 machine learning (ML) algorithms. The methodology included preprocessing to address missing values and standardizing data with the standard scaler. Hyperparameter optimization via random search enhanced model performance, and predictions were validated through 10-fold cross-validation to ensure reliability. Evaluation metrics such as mean squared error (MSE), mean absolute error (MAE), and the coefficient of determination (<i>R</i><sup>2</sup>) demonstrated the effectiveness of ridge-adaptive boosting (AdaBoost) regression, which achieved an outstanding <i>R</i><sup>2</sup> of 0.999 and an MAE of 0.001085. Visual tools, including <i>R</i><sup>2</sup> plots and learning curves, revealed insights into the model’s ability to explain variance and its learning process. A Friedman test confirmed significant performance differences across the models, solidifying ridge AdaBoost regression as the top performer. Comparative analysis with prior research highlighted the superior accuracy of this approach. The study contributes to artificial intelligence by advancing ML-based forecasting methods and supports engineering practices by optimizing PV system management. These results lay a foundation for future research and practical applications in enhancing the efficiency and sustainability of solar energy 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

Integrating local photovoltaic and meteorological data for enhanced forecasting precision using machine learning

  • 期刊论文

  • 发表日期：

    2026年7月22日

查看全文
  • Osman Dikmen,
  • Hasan Ulutas,
  • Erdem Elibol,
  • M Emin Sahin

摘要

Accurate energy forecasting is essential for improving the sustainability of the photovoltaic (PV) systems. This study utilizes six years of meteorological data from Duzce to predict solar energy output using 14 machine learning (ML) algorithms. The methodology included preprocessing to address missing values and standardizing data with the standard scaler. Hyperparameter optimization via random search enhanced model performance, and predictions were validated through 10-fold cross-validation to ensure reliability. Evaluation metrics such as mean squared error (MSE), mean absolute error (MAE), and the coefficient of determination (R2) demonstrated the effectiveness of ridge-adaptive boosting (AdaBoost) regression, which achieved an outstanding R2 of 0.999 and an MAE of 0.001085. Visual tools, including R2 plots and learning curves, revealed insights into the model’s ability to explain variance and its learning process. A Friedman test confirmed significant performance differences across the models, solidifying ridge AdaBoost regression as the top performer. Comparative analysis with prior research highlighted the superior accuracy of this approach. The study contributes to artificial intelligence by advancing ML-based forecasting methods and supports engineering practices by optimizing PV system management. These results lay a foundation for future research and practical applications in enhancing the efficiency and sustainability of solar energy systems.