This research explores the potential of solar power as an eco-friendly alternative to fossil fuels, focusing on Rajasthan, India. Using data from MERRA-2, researchers analysed solar irradiance and meteorological parameters to develop accurate prediction models. It is observed that for every 1 kW/m2 increase in shortwave downward solar irradiance, earth skin temperature rises by 3.52 °C. The temperature variable shows the highest correlation coefficient at 0.63, followed by wind with a correlation coefficient of 0.37, while surface pressure has a negative correlation coefficient of − 0.37. Various machine learning models were tested and compared for solar irradiance prediction, including Linear Regression, Random Forest, Decision Trees, Support Vector Machines, and Gradient Boosting. Results suggest, linear regression model demonstrates the best training and validation results, with a minimal difference of just 0.02 between the training and validation datasets. Performance of the model was assessed using Mean Squared Error, Root Mean Squared Error, Mean Absolute Error, and R-squared metrics. When looking at the overall performance of the various models, it’s clear that the Random Forest model is more effective, achieving a Root Mean Square Error of 0.64. This model outperforms the Linear Regression model by a notable 19.47%, highlighting its superior predictive abilities. Additionally, average error across all models stays below 20%, emphasizing the reliability and accuracy of the predictive analyses performed. This study provides a robust framework for predicting solar irradiance, crucial for efficient solar power planning and implementation. By enhancing our ability to forecast solar energy potential, this research could accelerate the adoption of renewable energy technologies worldwide. The success of the Random Forest model in this context highlights the power of advanced machine learning techniques in renewable energy applications. Ultimately, this study contributes to the global effort to transition towards cleaner energy sources, potentially influencing energy policies and investment decisions in solar power infrastructure.
Read full abstract