A Novel Method for Detection of Wind Turbine Blade Imbalance Based on Multi-Variable Spectrum Imaging and Convolutional Neural Network

This paper presents a novel method used for detecting blade imbalance occurring on wind turbines based on multi-variable spectrum imaging and convolutional neural network (CNN). The balance and imbalance conditions were simulated with the aid of commercial wind turbine design software Bladed. Simulation results including generator speed, generator torque and nacelle X acceleration were obtained and processed with Fourier transform to generate the combined spectral images, which were then fed to a CNN model in order to extract and learn the fault features. The effectiveness of the proposed method was validated through comparison with single variable CNN and fully-connected neural network. Results demonstrate that this method is capable of detecting aerodynamic imbalance and mass imbalance with high accuracy and high efficiency.