基于多视数据重构损失标定的叶片型面光学检测

Optical measurement of blade profile based on multi-view data reconstruction loss calibration

航发叶片型面检测是保证叶片加工质量、工作性能和使用寿命的关键环节。对此,该文开发一套基于线激光器的叶片型面检测平台;在完成线激光器与叶片相对位姿校准及Z向测量基准建立基础上,实现叶片型面特定截面处型线数据的多视采集;提出一种基于多视数据重构损失标定回转中心的型线重构方法。以两种典型燃气轮机导向叶片为检测对象,所提出方法获取的检测数据与三坐标测量机检测数据的对比结果表明,轮廓度检测平均绝对偏差在0.022 mm以内,最大标准偏差为0.0186 mm,最大RMS估计为0.0251 mm;特征参数检测最大偏差为0.0401mm。实验结果验证了所提出方法面向叶片型面检测的可行性。

Profile measurement for aero-engine blades is essential to ensure their processing quality,working performance,and service life.Therefore,a measurement platform based on laser scanning sensor is first developed for blade profile measurement.By calibrating the relative position between the blade and laser scanning sensor and establishing a measurement datum along the Z-direction,multi-view data acquisition of blade profile at specific cross-sections is achieved.Finally,a blade profile reconstruction method based on multi-view data reconstruction loss is proposed to calibrate rotational center.Compared with the measurement data obtained by coordinate measurement machine(CMM)on two typical gas turbine blades,the average absolute deviation on the measurement data obtained by the proposed method is within 0.022 mm,the maximum standard deviation is 0.0186 mm,the maximum root mean square(RMS)estimation is 0.0251 mm,and the maximum deviation of the main characteristic parameters is 0.0401 mm.The experimental results demonstrate the feasibility of the proposed method for blade profile measurement.