Wind Turbine Blades Damage Detection Based On Attention Enhanced Convolution Neural Network

In recent years,wind energy has been vigorously developed in our country,and the installed application of wind turbines has reduced emissions by about 150,000 tons per year,making wind energy an important renewable energy source.The detection of wind turbine blades can reduce the incidence of wind turbine operation accidents.In wind farms,unmanned aerial vehicle have been used to inspect wind turbine blades.Deep learning algorithms are widely used in the detection of wind turbine blade images.However,in deep learning algorithms,there are local characteristics of convolution kernels that cannot be used.Capturing the global context in the image leads to the loss of global information in the image.Aiming at this problem,this paper develops a wind turbine blade damage classification based on attentionenhanced convolution neural network.The main research contents are as follows:1.An image acquisition scheme of the wind turbine blades surface that can cooperate with manual inspection is proposed,and a highly feasible unmanned aerial vehicle acquisition is proposed as the blade image acquisition scheme.And an image data set suitable for the image recognition task of surface damage of installed wind turbine blades is established,which includes surface crack images of wind turbine blades,other damaged images and non-damaged images.2.A wind turbine blade damage detection model based on attention-enhanced convolution neural network(AECN)is proposed.The model achieves an average classification accuracy of96.67% in the classification and detection of crack damage,other damage and no damage,and has achieved good detection results.Compared with the convolutional neural network(CNN)and VGG(Visual Geometry Group)model,the effectiveness of this model is proved.The experimental results show that the algorithm can accurately capture the global information in the damage classification of wind turbine blades,and the classification accuracy is also higher.This algorithm provides a method basis for the extension and application of image recognition algorithm based on deep learning in the field of wind turbine blade surface damage detection.3.The software development of the image damage detection system is realized.Using Qt designer as the development environment,using the machine vision image processing software library as auxiliary development,and using the established wind turbine blades image data set,different levels of wind turbine blades damage image recognition and detection are realized.In this paper,a wind turbine blades image data set is established,and the image recognition and detection of wind turbine blades damage at different levels is realized,which provides a method basis for the promotion and application of the image recognition algorithm based on deep learning in the field of wind turbine blades surface detection.