Power Generation Forecasting of Wind Farms Using Machine Learning Algorithms

Abstract

AbstractIn this paper, long-term wind power generation forecasting is accomplished using five different types of machine learning (ML) algorithms. Forecasting is done based on wind power generation dataset consisting of various parameters for a three-year horizon. The proposed work using ML algorithms forecasts accurate wind power generation values with high accuracy and efficiency. The acquired results unveil the performance and appropriateness of chosen ML algorithms for wind power prediction. Proposed ML algorithm can be efficiently used prior to the installation of wind power plants in an unfamiliar place to determine the corresponding wind potential. Proposed techniques can also be employed for power pooling, where using the forecasts, power generation can be easily determined. Most appropriate ML algorithm with the best performance is identified and fine-tuned for further optimization and improved performance. A graphical user interface application is proposed for easy wind power forecasting and in extended applications.KeywordsMachine learning algorithmsWind energyWind power forecasting