基于拐点检测的椭圆弧叶型前后缘评定算法

叶片前后缘的形位对叶片气动性能有着显著影响,且其分割结果也严重影响特征参数的精密计算。现有的叶片前后缘分割算法大多基于拓延算法的实现与改进,但是拓延算法本身尚不能自适应叶型截面尺寸来分割点云,而后续的改进算法在轮廓线误差控制上仍有待加强。针对叶片截面线前后缘的点云精确分割问题,参照椭圆弧叶型的造型设计特点,提出了基于拐点检测的前后缘评定算法。通过对截面线点云旋转、截取、顺时针排序以及拐点检测处理,实现了在高精度测量条件下对叶片前后缘的精确分割。最后设计了对比实验,验证了本文算法可以定性地分割叶片前后缘,并在线轮廓误差上保持了较低水平;论证了在逐点拓延椭圆拟合过程中,从凹性点即开始影响线轮廓误差。

An evaluation algorithm of elliptic arc blade leading and trailing edges based on inflexion detection

The shape and position of blade leading and trailing edge have a significant impact on blade aerodynamic performance, and the result of segmentation seriously affects the precision calculation of characteristic parameters. Most existing segmentation algorithms of finding the location of leading and trailing edges are based on implementation and improvement of the extension algorithm; however, the extension algorithm could not adapt itself to the blade section size. What's more, subsequent improved algorithms on the contour error control remain to be strengthened. Aimed at segmentation problem of leading and trailing edges of the blade section line, an evaluation algorithm of leading and trailing edges based on inflexion detection is first proposed, with reference to the modelling of elliptic arc blade design characteristics. Then accurate segmentation of the blade leading and trailing edges is implemented under the condition of high-precision measurement, via point-clouds processing of rotation, interception, sorting and inflection point detection. And finally, a contrast experiment is designed to verify that this approach could split the blade leading and trailing edges qualitatively, keeping the contour error on a lower level, and demonstrate that the concavity points begin to influence line contour error in the process of ellipse fitting by the extension algorithm.