基于无人机巡检的风机叶片表面缺陷检测技术  

UAV-based inspection of wind turbine blade surface defects detection technology

作　　者：谭兴国[1,2] 张高明 TAN Xingguo;ZHANG Gaoming(School of Electrical Engineering and Automation,Henan Polytechnic University,Jiaozuo 454000,Henan,China;Hami Vocational and Technical College,Hami 839000,Xinjiang,China)

机构地区：[1]河南理工大学电气工程与自动化学院,河南焦作454000 [2]哈密职业技术学院,新疆哈密839000

出　　处：《电测与仪表》2025年第3期183-189,共7页Electrical Measurement & Instrumentation

基　　金：新疆维吾尔自治区自然科学基金(2022D01F46);哈密科技项目(hmkj202107);新疆维吾尔自治区人才发展专项(202102)。

摘　　要：“双碳”背景下大力发展新能源尤为重要,风力发电是重要的清洁能源,在新能源领域中风电规模也在一并扩大。随着风电机组的大型化,叶片受损概率也在增加,针对大型风力机叶片缺陷检测成本高、工作环境差等问题,文中提出了一种基于无人机采集图像和数字图像处理的风机叶片表面缺陷检测方法。针对无人机采集图像的特点,应用加权平均值实现灰度处理,再应用中值滤波实现图像降噪;并提出限制对比度自适应直方图均衡化(contrast limited adaptive histogram equalization,CLAHE)算法对图像进行增强,使目标区域和缺陷处细节更加清晰完整,提升了检测效率;通过图像前景分割及阈值处理等分离提取缺陷的特征信息,并由连通域进行框取,实现叶片表面的检测。通过引入性能评价指标平均交并比(mean intersection over union,MI-oU)来计算检测缺陷图像的准确率与误差率,经实验验证所提方法对砂眼、划痕、裂纹等典型叶片缺陷的检测准确率均在90%以上,尤其是裂纹缺陷的检测准确率可达到95%,从而验证了文章检测方法的有效性。Under the background of“double carbon”,it is particularly important to vigorously develop new energy.Wind power generation is an important clean energy,and the scale of wind power is also expanding in the field of new energy.With the increasing scale of wind turbines,the damage probability of blades is also increasing.Aiming at the problems of high cost and poor working environment of large-scale wind turbine blade defect detection,a wind turbine blade surface defect detection method based on UAV image acquisition and digital image processing is proposed in this paper.According to the characteristics of images collected by UAV,this paper adopts the weighted average meth-od to realize gray processing,and then,the median filtering is applied to realize image noise reduction;the image enhancement is realized by CLAHE algorithm,which makes the details of target area and defect more clear and complete,and improves the detection efficiency.The feature information of defect is separated and extracted through image foreground segmentation and threshold processing,and the connected domain is framed to realize the detection of blade surface.The accuracy and error rate of defect images is calculated and tested by introducing per-formance evaluation index MIoU.The experimental results show that the detection accuracy of the proposed method for typical blade defects such as trachoma,scratch and crack is above 90%,especially the detection accuracy of crack defects can reach 95%,which verifies the effectiveness and accuracy of the algorithm in blade detection.

关 键 词：风电叶片 叶片缺陷 对比度自适应直方图均衡化 缺陷检测 无人机 

分 类 号：TM614[电气工程—电力系统及自动化]