基于图像的风电叶片前缘雨蚀退化指标构建

针对风电叶片前缘加速雨蚀实验中试样质量损失指标实际应用场景单一的问题，提出了一种基于图像的雨蚀退化指标构建方法。采用标准化欧氏距离对图像灰度共生矩阵(Gray Level Co-Occurrence Matrix,GLCM)的多特征参数进行融合，构建退化指标(Degradation Indicator,DI)，并利用度量指标进行评估与筛选。以实验采集的140幅图像为例，将试样以不同侵蚀时长和不同半径区域为标签，分别构建了单一特征的DI和多特征融合的DI。结果表明，所构建的DI能直观地表示试样侵蚀的严重程度，对雨蚀阶段的区分准确率达95%；多特征融合能弥补单一特征的检测局限，使DI的适用性提高了5.5%。该DI构建方法验证了以图像为信号进行雨蚀检测的可行性。

Image-Based Degradation Indicator Construction of Leading Edge Rain Erosion For Wind Turbine Blade

Aiming at a single practical application scenario of the mass loss indicator in wind turbine blade leading edge accelerated rain erosion experiments,an image-based method for constructing rain erosion degradation indicators is proposed.A degradation indicator(DI) is constructed by fusing multiple features of the gray level co-occurrence matrix (GLCM) of images using the standardized Euclidean distance,and metrics are used for evaluating and screening different DI.Taking 140 images collected in the experiment as examples,a single feature DI and a multi-feature fusion DI are constructed by using different erosion durations and radius regions as labels.The results show that,the DI can visually represent the severity of erosion of the specimens and distinguish the erosion stages with 95% accuracy.Using multi-feature fusion can compensate for the detection deficiencies of single features and improve the suitability of DI by 5.5%.This method confirms that using images as signals for rain erosion detection is feasible.