大型海上风电机组叶片故障图像识别方法

Image recognition method for blade fault of large offshore wind turbine

针对海上风电机组叶片故障诊断建模时缺乏大量实际故障图像训练样本的问题,文章提出了一种基于小数据集的海上风电机组叶片故障图像识别方法。该方法针对风机叶片图像的叶片及其故障的颜色与形状特征,改进K均值聚类算法以实现叶片分割,设计自适应算法调整Canny算子参数以实现叶片表面早期故障区域的分割,使用K均值聚类算法提取故障的颜色和形状特征并设计相应的分类器以实现故障分类。仿真算例表明,该方法对于叶片表面早期故障的识别是有效的,可以在少量故障样本的基础上为海上风电机组叶片故障识别提供准确的诊断模型,提高了海上风电场的运维效率。

Aiming at the problem of lack of a large number of actual fault image training samples during the fault diagnosis and modeling of offshore wind turbine blades,an image recognition method for offshore wind turbine blade faults based on small data sets is proposed.In this method,the K-means clustering algorithm is improved to identify blade segmentation according to the color and shape characteristics of blades and their faults in wind turbine blade images,an adaptive algorithm is designed to adjust the Canny operator parameters to identify the segmentation of early fault areas on the blade surface,and the K-means clustering algorithm is used to extract the color and shape features of faults and design corresponding classifiers to achieve fault classification.Simulation examples show that this method is effective for the identification of early faults on the blade surface,and can provide an accurate diagnostic model for the blade fault identification of offshore wind turbines on the basis of a small number of fault samples,which can improve the operation and maintenance efficiency of offshore wind farms.