Scholarlink

The following paper investigates the usage and application of ensemble machine learning techniques for solar power prediction using complex and real-time inverter data. The data is obtained from a 370 KW solar power generation unit located in Chennai, Tamil Nadu, India. The data preprocessing and visualization show the data linearities, while highlighting upon the potential faults across the inverter generation units in the given system. The methodology proposed is an ensemble of Random Forest, Adaptive Boosting Regressor, and Extreme Gradient Boosting Regressor, along with weighted averaging as a combination technique. The results show that the final ensemble framework outperforms the individual models, by achieving a higher coefficient of determination (R2) of 0.966469 and lower values of errors—Root Mean Squared Error (RMSE) of 0.192171. The paper concludes that leveraging ensemble techniques offer flexibility and reliability in solar power forecasting due to its ability to capture linear and non-linear patterns in complex, real-time data.

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

Modeling Real-Time Solar Energy Prediction Using an Ensemble Learning Approach

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Anika Kamath,
  • S. Joyal Isac,
  • S. Amutha

摘要

The following paper investigates the usage and application of ensemble machine learning techniques for solar power prediction using complex and real-time inverter data. The data is obtained from a 370 KW solar power generation unit located in Chennai, Tamil Nadu, India. The data preprocessing and visualization show the data linearities, while highlighting upon the potential faults across the inverter generation units in the given system. The methodology proposed is an ensemble of Random Forest, Adaptive Boosting Regressor, and Extreme Gradient Boosting Regressor, along with weighted averaging as a combination technique. The results show that the final ensemble framework outperforms the individual models, by achieving a higher coefficient of determination (R2) of 0.966469 and lower values of errors—Root Mean Squared Error (RMSE) of 0.192171. The paper concludes that leveraging ensemble techniques offer flexibility and reliability in solar power forecasting due to its ability to capture linear and non-linear patterns in complex, real-time data.