FORECASTING OF POWER FOR SOLAR PV AND WIND ENERGY SYSTEM

Abstract

Renewable energy sources like solar and wind are essential for reducing greenhouse gas emissions and promoting sustainable power generation compared to conventional fossil fuel sources. However, the intermittent nature of these renewable energy sources presents challenges in maintaining a stable power supply. Accurate forecasting is crucial for managing these challenges by predicting the future output of wind and solar energy sources. In this report, we explore various forecasting techniques used for solar PV and wind energy, including statistical models and machine learning algorithms. We conducted both univariate and multivariate analyses using four different machine learning models: random forest, support vector machines, k-nearest neighbours and artificial neural networks to forecast Global Horizontal Irradiance (GHI) and wind speed for two locations, Roorkee (India) and Belgrade (Serbia). We discuss the advantages and limitations of each method and their applicability in different scenarios. Additionally, we analyse the impact of weather variables on forecasting accuracy. Our findings suggest that machine learning algorithms, particularly random forest and artificial neural networks, can significantly improve the accuracy of GHI and wind speed forecasting. These results provide valuable insights into designing efficient and reliable renewable energy systems for a sustainable future.