Scholarlink

<p>This study proposes a polynomial regression model with Ridge regularization for forecasting the combined renewable energy generation (solar and wind) in Dhaka City. The model was developed using over a decade of meteorological data (2014–2023), capturing non-linear interactions among key predictors such as temperature, wind speed, and solar irradiance. Data preprocessing included normalization and outlier handling. Model performance was evaluated using standard metrics—mean squared error (MSE), mean absolute error (MAE), root mean square error (RMSE), and coefficient of determination (R<sup>2</sup>)—with 80/20 train-test splitting and 05-fold cross-validation ensuring generalizability. The Ridge-regularized model achieved an R<sup>2</sup> of 0.99 on the test data, with very low normalized error values (MSE ≈ 2.42 × 10<sup>−8</sup>, MAE ≈ 7.59 × 10<sup>−5</sup>), corresponding to an RMSE of approximately 1.55 × 10<sup>−4</sup>. While these error values are small due to normalization, the R<sup>2</sup> score provides a robust, scale-independent measure of performance. The high accuracy is supported by the strong correlations among features and the model’s ability to generalize effectively across cross-validation folds. By integrating wind and solar forecasting under a unified framework and applying Ridge regularization to manage multicollinearity, this work offers a reliable and interpretable approach for renewable energy planning in rapidly urbanizing regions.</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

Polynomial regression approach with ridge regularization for forecasting wind and solar energy in Dhaka City

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月26日

查看全文
  • Mohammad LitonHossain,
  • S. M. Nasif Shams,
  • Saeed Mahmud Ullah

摘要

This study proposes a polynomial regression model with Ridge regularization for forecasting the combined renewable energy generation (solar and wind) in Dhaka City. The model was developed using over a decade of meteorological data (2014–2023), capturing non-linear interactions among key predictors such as temperature, wind speed, and solar irradiance. Data preprocessing included normalization and outlier handling. Model performance was evaluated using standard metrics—mean squared error (MSE), mean absolute error (MAE), root mean square error (RMSE), and coefficient of determination (R2)—with 80/20 train-test splitting and 05-fold cross-validation ensuring generalizability. The Ridge-regularized model achieved an R2 of 0.99 on the test data, with very low normalized error values (MSE ≈ 2.42 × 10−8, MAE ≈ 7.59 × 10−5), corresponding to an RMSE of approximately 1.55 × 10−4. While these error values are small due to normalization, the R2 score provides a robust, scale-independent measure of performance. The high accuracy is supported by the strong correlations among features and the model’s ability to generalize effectively across cross-validation folds. By integrating wind and solar forecasting under a unified framework and applying Ridge regularization to manage multicollinearity, this work offers a reliable and interpretable approach for renewable energy planning in rapidly urbanizing regions.