一种基于深度学习的丝杠螺母内滚道型面稀疏重建方法
By interpolating and upsampling sparse measurement point clouds using helical spatial decoupling transformation and deep learning models, the problem of low detection efficiency of inner raceway in ball nuts is solved, achieving efficient and accurate reconstruction of inner raceway profile and improving the engineering practicality of the detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- YANQIHU BASIC MFG TECH RES INST (BEIJING) CO LTD
- Filing Date
- 2026-06-15
- Publication Date
- 2026-07-17
AI Technical Summary
In the existing technology, the detection method of the inner raceway of the ball nut cannot quickly obtain a complete and dense three-dimensional surface morphology, resulting in low detection efficiency and difficulty in achieving a comprehensive and efficient evaluation of the inner surface morphology of the entire deep hole spiral.
A spiral spatial decoupling transformation and a deep learning model are used to intelligently interpolate and upsample sparse measurement point clouds. The predicted point cloud is output through the reconstruction model of the sparse sample set. Combined with the loss function and inverse matrix operation, a dense point cloud set is generated to achieve rapid reconstruction of the inner roller track surface.
It improves the efficiency and accuracy of inner raceway measurement, and the generated dense point cloud has higher physical realism and geometric accuracy on key functional surfaces, significantly enhancing the engineering practicality and robustness of the inspection.
Smart Images
Figure CN122415902A_ABST