Investigation of icing causes on wind turbine rotor blades using machine learning models, minimalistic input data and a full-factorial design

Abstract

Listen

Translate article

Ice formation on rotor blades of wind turbines cause significant downtimes in cold regions. Existing ice sensors can detect, but cannot predict ice. Therefore, our research aims to develop an ice prediction system based on historical and forecasted data. In a first step, the detection process is analyzed in this research paper. Two machine learning models are trained using minimalistic input parameters. A full-factorial experiment design is performed for the models, using the fl-score as the response variable. The most significant input parameter was the external temperature for both models.