Detection And Analysis Of Wind Turbine Blades Surface Faults Based On UAV-Taken Images

Wind energy is one of the most important renewable energy sources in modern society.Wind turbine blades are the main devices for capturing wind energy and converting it into electrical energy.Because all kinds of blades installed in commercial wind farms in China are exposed to harsh natural environment,withstand high-frequency mechanical loads,and undergo various extreme weather erosion,the blades are prone to various faults.Coupled with some unreasonable design process and material selection factors,if these faults are not handled in time,they will soon cause all kind of accidents in a short period of time,which will bring great economic losses to the operation of the entire wind farm.Timely detection and analysis of wind turbine blade surface failure is particularly important.At present,the surface faults of blades in China are mainly divided into two categories: crack and corrosive surface.In the research,UAV is used to collect all kinds of fault images on the surface of blades.Through the corresponding image processing algorithms,the collected images are processed as follows: de-motion blur,noise elimination,image enhancement,image segmentation,feature extraction and fault classification;areas and categories of blade surface faults are detected in time.Compared with the traditional manual detection method,it is more convenient,faster and safer.Compared with the various sensor detection methods that are popular nowadays,the installation and power supply of the sensor are omitted.The main research work of this thesis is as follows:1、The current status of global wind energy development is analyzed.The various faults and causes of wind turbine blade surface are introduced in detail.The advantages and disadvantages of various detection methodsare described by comparison.The detection and analysis scheme of wind turbine blade surface fault based on image acquired by drone Been proposed;2、The environmental characteristics of the wind turbine operation are carefully considered.The main structure of the UAV and the flight control principle are carefully analyzed.A set of fan blade fault image acquisition scheme is developed.Combined with the imaging characteristics of the UAV,a set of acquisition is designed.Image sharpening pre-processing research scheme provides guarantee for subsequent image analysis;3、The fault image geometry,texture and gray scale are extracted,and a 12-dimensional feature pool is established.The characteristic dimension reduction study is carried out in combination with the ‘dimensional disaster’ theory,and verified by conventional classifier;4、A partial binary tree classification system based on twin support vector machine is constructed,which solves the problem of partial fault classification.Under the same conditions,the classification accuracy is improved to 92.5%,and the classification efficiency is greatly improved.