[2C3] Acoustic emission-based health monitoring of composite wind turbine blades
S Sikdar¹, S Banerjee² and R Mishra¹
¹University of Huddersfield, UK
²Indian Institute of Technology Bombay, India
Condition monitoring of wind turbines is vital to ensure their efficient and reliable operation, reducing downtime and maintenance costs while maximising energy generation. This research presents an innovative and concise method for acoustic emission (AE)-based intelligent structural health monitoring (SHM) of composite wind turbine blades using a hybrid machine learning (HML) model. The model uses damage-induced AE signals from experiments. The proposed methodology involves real-time acquisition of damage-induced AE signals from laboratory-based glass fibre-reinforced composite wind turbine blades, processing of the registered AE signals to time-frequency spectrograms and using them for training, validation and testing of a designed HML model for autonomous damage source identification. The HML model uses a designed convolutional neural network for feature extraction and a support vector machine for classification/identification using the extracted features. This integration offers a robust in-service SHM system for the wind turbine blades compared to traditional methods, making it more practical for wind farm environments by deploying wireless sensors in future. The findings show potential towards enhancing wind turbine design to increase reliability and efficiency, contributing to an eco-friendly sustainable energy generation.
