LAUSR

Abstract In recent years, with the successive development of mountain wind power resources in the high humidity regions, the problem of ice covering on wind turbine blades has become increasingly prominent. Wind turbine blade icing will affect blade aerodynamic performance and reduce power generation. In this paper, a method for measuring the ice thickness of wind turbine blades based on edge detection was proposed. As the icing images of blades have characteristics like low contrast, unclearness and blurred edges, the images were preprocessed by methods like defogging, gray-scale transformation, histogram correction, Gaussian filtering, and Laplace edge enhancement. Aiming at the characteristics of wind turbine icing images with many noises and many small edges, a multiscale wavelet edge detection algorithm was used to detect the blade icing images. Based on the pixel coordinates of the ice edge and the edge of the blade, through the pre-calibration, a method for calculating the ice thickness of the leading edge of the blade based on the pixel coordinate difference /real length ratio was proposed. Multiple measurement tests were performed in the natural icing environment. The average ice thickness measurement error value of short distance (0.5 m)/long distance (35 m) is 0.31 mm/8.9 mm, and the average ice thickness measurement error rate is 2.62% /6.0%.