Wind Energy Analysis and Forecast using Machine Learning

Abstract

Better prediction tools for future solar and  wind power are crucial to reducing the requirement for  controlling energy associated with the conventional power  facilities. For optimal power grid integrating of highly  variable wind power output, a strong forecast is extremely  crucial. In this part, we concentration on wind power for  the near run projections and conduct a wind unification  study in the western United States using data from the  National Research Conducted by the university (NREL).  Our approach derives functional connections directly from  data, unlike physical systems that rely on exceedingly  difficult differential calculus. By recasting the prediction  problem as a regression problem, we investigate several  regression methodologies such as regression models, k nearest strangers, and regression algorithms. In our testing,  we look at projections for specific machines along with  power from the wind parks, proving that a classification  algorithm for predicting short-term electricity generation is  feasible.