IoT based Architectural Framework for Wind Speed and Solar Radiation Prediction

Abstract

Wind speed and solar radiation are the two important renewable energy resources. Both environmental parameters show non-linear behavior. It is necessary to predict these parameters for more accurate results. The study indicates that good quality prediction using the machine learning methodology can predict correct data such as temperature, relative moisture, solar radiations, rain, and wind speed. ANN model gives accurate results for non-linear processes. This paper compares advantages and limitations of different ANN models such as Bayesian Regularization (BR), Resilient Back Propagation (RP), Levenberg Marquardt (LM), Polak-Ribiére update gradient (CGP) and OSS (one step secant) gradient to enhance the accuracy of the wind speed and the predicted solar radiation. The analysis shows that the Levenberg-Marquardt and Bayesian Regularization algorithms are more resilient and efficient to predict highly non-linear characteristics, such as solar radiation and wind speed. After collecting accurate results, IoT is the better platform utilized for real-time data analysis to increase energy efficiency, dependability, detection of failures, and production optimization.