Solar Irradiance Prediction for Agricultural Applications Using Hybrid Machine Learning

Abstract

The accurate prediction of solar irradiance becomes crucial for effective management of solar-powered agricultural systems, such as irrigation pumps, running of greenhouses, crop drying, livestock sheds, farm energy storage, etc. Nonetheless, solar irradiance is very fluctuant due to variations in cloudiness, temperature, humidity, wind speed, season, etc. The aim of the present study is to introduce a hybrid machine learning system to accurately predict the solar irradiance for agricultural applications. The proposed approach integrates complementary machine learning algorithms that are able to capture linear and nonlinear relationships in the historical weather and irradiance data. The inputs to the model are temperature, relative humidity, wind speed, cloud cover, rainfall, time, and past solar irradiance readings. Data preprocessing, feature selection and hyperparameter optimization are used to enhance the reliability of the model and to minimize prediction errors. The mean absolute error, root mean square error, mean absolute percentage error, and coefficient of determination are used to measure the performance of the hybrid model. The proposed framework is likely to yield more accurate and stable predictions than the individual machine learning models. Accurate solar irradiance prediction can help optimize irrigation scheduling, battery charging, energy distribution and agricultural equipment operation. The research helps to advance intelligent, low cost and sustainable energy management systems for precision agriculture and climate-smart farming.