Method and apparatus for identifying edges of wind turbine components

Abstract

A method and apparatus for identifying the edge of a wind turbine component are provided. The method includes: acquiring an image including a component to be identified; inputting the acquired image to a convolutional neural network model trained based on a sample set to obtain an image indicating an edge of the component to be identified, wherein the convolution The neural network model includes N convolutional stages, the kth convolutional stage includes Mk convolutional layers, N is an integer greater than 1, k is an integer greater than 0 and less than or equal to N, and Mk is an integer greater than 1, where The input layer of the above convolutional neural network model is connected to the first convolution stage, starting from the first convolution stage, each convolution stage is connected to the next convolution stage through the pooling layer, and the N convolution stages are All convolutional layers included are also connected to the output layer through a specific network structure. According to the method and device, the edge of the wind power generator components can be recognized quickly and accurately from the image.

Description

technical field [0001] The present invention generally relates to the field of wind power generation, and more specifically relates to a method and device for identifying edges of wind power generator components. Background technique [0002] The environment in which wind turbines are located is complex and changeable, and wind turbines can be found in various geographical environments such as forests, grasslands, deserts, Gobi, hills, coastal areas, and seas. Wind turbines under various landform conditions often have various degrees of appearance damage, which affects power generation. [0003] Image recognition technology has advantages in identifying the appearance damage of wind turbine generators. It can quickly and accurately identify various appearance damages such as icing, cracks, cracks, and fouling of blades of wind turbine generators. However, due to the complex and changeable geographical environment of wind turbines, the background noise in the collected image...

AI Technical Summary

Problems solved by technology

However, due to the complex and changeable geographical environment of wind turbines, the background noise in the collected images of wind turbine components can easily lead to the failure of appearance damage recognition

For example, for the identification of icing on wind turbine blades, the background noise in the captured blade images (such as clouds in the sky, light spots, etc.) has certain similarities with the characteristics of icing on the blades, and it is easy to be misidentified as icing. This leads to identification errors; for the identification of blade cracks, the background noise (for example, trees, rocks, distant wind turbines, etc.) Blade cracks lead to identification errors

Images