Elements of Nature Optimized into Smart Energy Grids using Machine Learning
Abstract
Accurate forecasting of renewable energy sources plays a key role in their integration into the grid. We propose the ability of machine learning algorithms to predict solar radiation based on the input parameters such as humidity, Wind direction, temperature, pressure humidity and radiation. The machine learning algorithms gets under reinforcement carefully observing past patterns and seasonality of the radiation. An assortment of machine learning algorithms used such as Linear Regression, Lasso Regression, Random Forest Regression and Support Vector Machine to understand the nature of the seasonal data. The evaluative accuracy metrics used here is mean absolute error and cross validation score. The most efficient machine learning algorithm in the due scenario turned out to be Hyper-Parametrized Random Forest Algorithm with a whooping cross validation score of approximately 72%. The success of the corresponding algorithm is attributable mainly to its ability to capture the diurnal cycle more effectively than other methods.