A Review of Machine Learning Approaches for Prediction of Icing on Wind Turbines

Icing on the blades of wind turbines which operate in cold climates is a major concern due to its impact on their power generation capacity. Prediction of icing before it occurs would allow taking evasive operational or control actions to reduce power losses. This paper presents a review of machine learning approaches that have appeared in the literature to predict icing on wind turbines. The focus in this paper is placed on those prediction approaches that utilize SCADA data or those approaches that do not require the installation of additional sensors on the turbine. By considering these approaches in one place in this review, a number of observations are made for consideration in future research towards achieving verifiable accuracies for prediction of icing.