Scholarlink

<p>Predicting solar radiation hourly for a whole month is crucial for optimizing solar energy production and storage. While several solar radiation forecasting studies exist, their ability to generalize to fresh data remains limited, highlighting the need for more reliable prediction methods. This study proposes a hybrid Coati Optimization Algorithm with an Artificial Neural Network model (COA-ANN) for solar radiation forecasting. Feature selection using Pearson correlation and probability methods helped identify the most significant weather parameters for efficient solar radiation prediction. Using data collected from VIT University in Vellore, Tamil Nadu, India, to evaluate the accuracy of COA -ANN, the model’s accuracy was assessed using statistical benchmarks and compared against ANN, PSO- ANN, and GWO-ANN models. Comparative analysis revealed that our proposed COA-ANN model achieved the lowest error metrics of RMSE (35.6009), MAE (2.6557), MAPE (0.04558), and highest coefficient of determination among all tested approaches with an R² of 97.244%, demonstrating its accuracy in predicting solar radiation under varied meteorological conditions.</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

Enhancing solar radiation forecasting via feature selection in hybrid artificial neural networks: a comparative study with coati optimization

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月29日

查看全文
  • T. K. Nida Fariz,
  • S. Sharief Basha

摘要

Predicting solar radiation hourly for a whole month is crucial for optimizing solar energy production and storage. While several solar radiation forecasting studies exist, their ability to generalize to fresh data remains limited, highlighting the need for more reliable prediction methods. This study proposes a hybrid Coati Optimization Algorithm with an Artificial Neural Network model (COA-ANN) for solar radiation forecasting. Feature selection using Pearson correlation and probability methods helped identify the most significant weather parameters for efficient solar radiation prediction. Using data collected from VIT University in Vellore, Tamil Nadu, India, to evaluate the accuracy of COA -ANN, the model’s accuracy was assessed using statistical benchmarks and compared against ANN, PSO- ANN, and GWO-ANN models. Comparative analysis revealed that our proposed COA-ANN model achieved the lowest error metrics of RMSE (35.6009), MAE (2.6557), MAPE (0.04558), and highest coefficient of determination among all tested approaches with an R² of 97.244%, demonstrating its accuracy in predicting solar radiation under varied meteorological conditions.