Quantification of Strategic Reserve for a Wind Farm Based on a Linear Regression Machine Learning Scheme: A Case Study of Great Kei Local Municipality, South Africa

Abstract

Inherent stochastic nature of generating electric power from wind power poses a challenge to short-term electricity markets. In this paper we collected expected electricity generation for wind power and weather data at one hour interval over a period of one year at Great Kei Local Municipality in South Africa. A linear regression machine learning scheme is subsequently used to predict expected available wind power as well as the quantification of strategic reserve. The results show that there is good degree of alignment between the observed and predicted wind power. The scheme is also able to quantify strategic reserve needed by wind power producers (WPP) in order to mitigate losses during participation in short-term electricity market. A strategic reserve margin needed for a wind farm with an installed capacity of 21 MW , is estimated to be ± 1.344 MW .