查看文章

Icing is a complex problem facing wind turbines operating in Nordic climates. It is affected by various instantaneously fluctuating parameters. Due to the deformation of blades' airfoil on account of icing, a significant drop in aerodynamic performance brings turbines to lose much of their productivity. Modelling and simulation became indispensable tools to estimate the effect of icing on the operation of wind turbines. However, the analysis of the iced airfoils via simulation is not practical for real-time ice prediction. This paper uses a combined analytical-CFD simulation approach to rapidly estimate the aerodynamic parameters of iced airfoils. The method is used to create a database of aerodynamic losses for several scenarios of weather conditions. The results have been processed by a neural network optimization algorithm to predict the aerodynamic losses of iced airfoils under varying scenarios of icing conditions. The ANN analysis results demonstrated consistency between the different scenarios of the database. They also highlighted the important influence of certain simulation parameters, such as the liquid water content and the angle of attack. Further investigation is recommended to determine correlations between the relevant parameters.