SOLAR FORECASTING FOR CROP CULTIVATION IN PESHAWAR AND MARDAN REGIONS: A COMPARATIVE ARTIFICIAL INTELLIGENCE ANALYSIS

Abstract

In the regions of Peshawar and Mardan, where sunlight has a major effect on crop production, precise solar forecasting facilitates optimization in agricultural activities. Predictive analysis was performed on a meteorological dataset employing Artificial Intelligence (AI) methods like Long Short-Term Memory (LSTM), Random Forest (RF), and Support Vector Regression (SVR). These methods are used to analyze the data and forecast solar irradiance. Performance of the models is evaluated based on measures like Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and R² values. Experimental results show that LSTM outperforms RF and SVR in terms of both precision and accuracy and achieves an RMSE of 12.45 W/m² for Peshawar and 14.32 W/m² for Mardan. This research aims to enhance solar energy prediction with AI to increase precision in agriculture for semi-arid areas.