| In order to implement the "30.60" carbon reduction target,China’s clean energy has been developed rapidly.Wind energy has been widely promoted because of its cleanliness,environmental friendliness and abundant reserves.About 75% of China’s wind farms are concentrated in the three northern regions.In the spring and winter season,icing is very likely to occur,and blade icing seriously affects the economy and safety of wind turbine operation and reduces the effective running time.Therefore,it is of great practical significance to carry out online monitoring technology research of wind turbine blade ice thickness.1.According to the characteristics of wind turbine blade ice cover,image pre-processing method are studied.The defects of insufficient image contrast caused by lighting changes are improved by linear grayscale transformation.The non-local mean filtering algorithm based on integral image is determined by comparing and analyzing different types of denoising algorithms,and the filtering parameters of this algorithm are optimized by adding Gaussian noise with different standard deviations.2.Due to the complex image background,for the defects of high time complexity and low accuracy of traditional OSTU algorithm,the weight function of particle swarm algorithm is improved.And the separation of target blade and background is better accomplished by using OSTU combined with particle swarm algorithm,which improves the accuracy and efficiency of target blade and background separation.3.The leaf edges extracted by the traditional edge detection algorithm often have poor connectivity and many pseudo-edges.By constructing structural elements of different sizes and scales,the traditional morphological edge detection algorithm is improved,and the accuracy of the improved algorithm is verified through detection comparison.4.Using picture distortion correction and pixel coordinate change,the calculation method of ice cover thickness is obtained.5.Experiment of wind turbine blade ice cover thickness measurement is carried out.The images of wind turbine blade model before and after ice cover were collected,and the ice cover thickness was obtained by image pre-processing,blade edge detection,distortion correction and coordinate transformation.The deviation of ice cover thickness measurement was less than 2mm,and the relative error was within 8%. |