A Categorisation Wind Power Forecasting Methodologies, Highlighting Emerging Short-Term Forecasting Methods

Global wind power capacity has seen historic growth in the last decade and is anticipated to increase rapidly in the future. As such the ability to accurately forecast wind power, at all timescales, is important to both central network operators and individual wind power producers. This paper provides an overview of wind forecasting methodologies, focusing on their timescales of applicability, along with their performance, advantages, and shortcomings. This paper's contribution is in analysing, synthesising, and critiquing a large amount of wind power forecasting research, thereby clarifying the state of current knowledge and making it more accessible to researchers and decisions makers within the energy industry. To highlight rapidly developing areas within wind power forecasting, a novel distinction between fundamental and emerging forecasting methodologies is made. Fundamental methodologies include time series based and artificial intelligence methods; emerging methodologies, in turn, are built on fundamental methodologies and include spatial-temporal forecasting, probabilistic forecasting, combined forecasting methods, and the incorporation of exogenous atmospheric variables into forecasts. Furthermore, as very little research has been done on short-term forecasting specific to South Africa, the focus of the review is on statistical forecasting methods which are well suited to the short-term time horizon and are computationally inexpensive. It is anticipated that the insights provided by this structured review will contribute to the development of effective forecasting frameworks necessitated by the increase of wind power both globally and specific to the South African network.