Topological Data Analysis Applied to Wind Turbine Vibration Spectra for Blade Icing Detection

Ice build-up on wind turbine blades is a significant issue, leading to operational risks and reduced efficiency. Traditional detection methods, such as visual inspection, power curve analysis or specialised sensors, are often slow, inefficient, or costly. This paper proposes an approach using 0-dimensional persistence homology from topological data analysis (TDA) on tower and blade vibration spectra to extract key features representing the lifespan of the sub-level sets to formulate a clearer supervised learning problem. Persistence diagrams are embedded in persistence images and rank functions, enabling ice detection through convolutional neural networks (CNN), and functional principal component analysis (FPCA) with support vector machines (SVM). This approach shows promise for reducing equipment and maintenance costs, leading to more efficient blade monitoring and maintenance processes.