| 夏桂书1 ，莫杜衡 1 ，魏永超 2 ，刘家伟 3 ，朱泓超 3.基于单视角涡轮叶片叶尖点云的增材修复路径规划[J].航空发动机,2025,51(2):169-174 |
| 基于单视角涡轮叶片叶尖点云的增材修复路径规划 |
| Additive Repair Path Planning Method Based on Single-View Turbine Blade Tip Point Cloud |
| |
| DOI： |
| 中文关键词: 涡轮叶片 叶尖损伤 再制造 增材修复 路径规划 点云 偏差分析 航空发动机 |
| 英文关键词:turbine blade blade tip damage remanufacturing additive repair path planning point cloud deviation analysis aeroengine |
| 基金项目:四川省科技厅重点研发项目（2022YFG0356）、西藏科技厅重点研发计划（XZ202101ZY0017G）、中国民用航
空飞行学院科研基金（J2020-040，CJ2020-01）资助 |
|
| 摘要点击次数: 172 |
| 全文下载次数: 116 |
| 中文摘要: |
| 针对当前航空发动机叶片增材路径生成过程繁琐、误差较大的问题，提出一种基于单视角涡轮叶片叶尖点云的完整规
划方法用于增材修复中的机械臂轨迹规划。现有的方法引入了曲面重构、点云拼接步骤，过程繁琐且引入额外误差；新方法基于
结构光3维扫描仪的单视角扫描结果，根据单视角点云的噪声分布采用优化滤波算法提取轮廓、统一姿态，投影轮廓在2维平面进
行分段拟合并平滑连接涡轮叶片叶尖的中心线路径点，并将路径点还原至3维初始位置。通过试验对不同姿态下2种不同类型的
涡轮叶片进行修复路径规划，将得到的路径点与模型进行偏差分析。结果表明：相较于曲面重构方法，采用新方法得到的标准偏
差和RMS估计值减小均超过了90%；相较于完整点云方法，采用新方法得到的标准偏差和RMS估计值也至少减小了4.6%，证明了
新方法的有效性与精确性均较高。 |
| 英文摘要: |
| In response to the problems of complex process and large errors in the current additive path generation method for
aeroengine blades, a comprehensive planning method based on single-view point cloud data of turbine blade tips was proposed for robotic
trajectory planning in additive repair. Existing methods involve surface reconstruction and point cloud stitching steps, which are cumber?
some and introduce additional errors. The new method leverages the results of single-view scans from a structured light 3D scanner. Based
on the noise distribution in the single-view point cloud, an optimized filtering algorithm was adopted to extract the contour, unify the
posture, project the contour onto a 2D plane for segmentation fitting, smoothly connect the central path points of the turbine blade tip, and
restore the path points to their original 3D positions. Repair path planning for two different types of turbine blades in varying postures was
conducted, and deviation analyses were performed between the obtained path points and the model. The results show that compared to
surface reconstruction methods, the standard deviation and RMS error estimates obtained by the new method are reduced by more than
90%; compared to the complete point cloud method, the standard deviation and RMS error estimates are also reduced by at least 4.6%,
demonstrating the algorithm's effectiveness and accuracy. |
| 查看全文 查看/发表评论 下载PDF阅读器 |