Study On Spectral Characteristics And Damage Classification Of Wind Turbine Blade

In recent years,wind energy,as a new type of renewable and clean energy,plays an important role in China’s electric energy structure because of its environmental protection and low cost.As an important tool for converting wind energy into electrical energy,blades are a vital part of wind turbines,and their integrity and operation are the key to ensuring the reliability and performance of wind turbine systems.In order to ensure the stable operation of the wind turbine system,it is necessary to regularly detect and maintain blade damage to prevent accidents,which is of great significance to the operation of wind turbines and the development of the wind power industry.At present,wind farm detection adopts telescope artificial visual inspection,there are shortcomings such as poor detection accuracy and low efficiency,and the commonly used ultrasonic,acoustic emission,infrared thermal imaging and other detection methods in industry have limitations in application due to the need to contact blades.Recently,methods such as damage detection based on UAV blade image acquisition have attracted widespread attention.Based on the automatic inspection and acquisition of blade images by UAV,combined with material damage mechanism analysis and image processing technology,this dissertation studies the damage location,identification and classification of wind turbine blade images,and realizes blade damage detection in wind farms.The main research contents of this dissertation include:(1)The structure of wind turbine and the material of wind turbine blades were analyzed.Through Fourier infrared spectroscopy,the material of wind turbine blades was qualitatively analyzed.It lays the foundation for the subsequent analysis of blade damage characteristics.(2)The damage mechanism of blade materials was analyzed.The finite element method is used to model the wind turbine blades,and the blade forces are analyzed based on the simulation of the temperature and wind speed conditions of the wind field,and the fatigue characteristics of the blades and the formation mechanism of blade damage are analyzed.(3)Build a leaf damage image dataset.Collect onshore and offshore wind turbine blade images from Sanmenxia,Inner Mongolia,Yancheng and other places,screen blade damage image data based on blade material damage mechanism,and make blade damage image datasets,which can be used to train deep learning models to identify and classify different types of blade damage.(4)Blade damage classification is realized based on deep learning technology.Combined with the blade damage dataset,the wind turbine blade damage image is preprocessed,and then the deep learning model is trained based on the YOLOv5 model and the image classification is realized.The algorithm based on the combination of image preprocessing and YOLOv5 recognition proposed in this dissertation is compared with Faster-RCNN,SSD and other algorithms to verify the effectiveness of recognition accuracy and speed.A damage recognition software based on deep learning algorithm is developed to provide an operation detection platform for actual wind farms,and generate blade damage detection reports for wind power operation and maintenance enterprises.