引用本文: 马宝, 马晓宝, 兰媛, 任玉成, 刘云飞, 马彦芝, 王荣军, 江连运. 翘曲板材三维重构及智能化压平的建模研究[J]. 工程科学学报, 2024, 46(5): 875-886.
MA Bao, MA Xiaobao, LAN Yuan, REN Yucheng, LIU Yunfei, MA Yanzhi, WANG Rongjun, JIANG Lianyun. Modeling research on three-dimensional reconstruction and intelligent flattening of warped plate[J]. Chinese Journal of Engineering, 2024, 46(5): 875-886.
 Citation: MA Bao, MA Xiaobao, LAN Yuan, REN Yucheng, LIU Yunfei, MA Yanzhi, WANG Rongjun, JIANG Lianyun. Modeling research on three-dimensional reconstruction and intelligent flattening of warped plate[J]. Chinese Journal of Engineering, 2024, 46(5): 875-886.

Modeling research on three-dimensional reconstruction and intelligent flattening of warped plate

• 摘要: 为提高板材压平质量和效率，实现智能化压平，本文采用结构光相机获取翘曲板材点云数据，进行点云分割、点云去噪和点云精简等预处理，利用最小二乘法拟合重构了翘曲板材的三维曲面. 采用微分几何的曲面理论计算得到板材的翘曲曲率，提出基于三维曲面特征的翘曲板材垫铁支点和压下点的优化方法，根据三点弯曲矫平理论得到了压力下和压下位移. 翘曲板材三维轮廓的重构方法和翘曲曲率计算模型确定了垫铁位置和压下位置点优化原则，为翘曲板材智能化压平提供了理论模型. 理论和实验的结果显示，重构的三维曲面与实际结果吻合，基于重构三维曲面数据计算的压平力与实测结果偏差大约为2.21%，初始平直度17.2 mm·m–1，压平矫直后减小为3.28 mm·m–1，结果准确，方法可行.

Abstract: To enhance the flattening quality and efficiency of plates and realize intelligent flattening, in this work, machine vision is utilized in place of manual recognition to realize accurate recognition of the three-dimensional (3D) contour of warped plates. Modeling research on the relevant processes of point cloud recognition, surface fitting, and flattening process is conducted. A structured light camera is utilized to obtain the point cloud data of the 3D contour of warped plates, and the recognition accuracy is found to meet the requirements of the flattening of the warped plate. Preprocessing of the point cloud data is conducted. The ground point cloud is segmented from the point cloud of the warped plate based on a shape model method. The inner outline of the ground point cloud is extracted, and afterward, the warped plate at any position is converted into a new pose, with the plate center serving as the origin of the camera coordinate system and the plate edge being parallel to the coordinate axis. It provides a foundation for acquiring the warped plate distribution in the coordinate system of the flattening machine, but it also facilitates subsequent point cloud denoising, which is conducted using conditional filters and distance-based techniques. Finally, the point cloud is simplified to enhance the computational efficiency using the principle of sparse sampling points. The 3D surface of the warped plate is reconstructed using the least-squares method. The curvature of the warped plate is computed using the surface theory in differential geometry, and the positions of the warped plate’s pad and pressing points are optimized based on the 3D surface features. Based on the three-point bending leveling theory, the screw-down force and displacement of warped plate are obtained. In comparison to the conventional modeling research of the flattening process based on the two-dimensional warping contour, we establish a reconstruction method of the warped plate’ 3D contour and a curvature model and determine the optimization principles for the positions of the pad plate and pressing point thereby providing theoretical models for intelligent flattening of plates. A finite element model of the flattening of the warped plate demonstrates that the deviation between the simulation and theoretical results for the flattening force is less than 5.4%, and the residual warping height is about 1 mm, indicating the reliability of the theoretical model. The deviation between the simulation and theoretical screw-down force is about 2.21%, and the unevenness decreased from 17.2 to 3.28 mm·m−1, demonstrating that the models and methods are relatively accurate and feasible.

/

• 分享
• 用微信扫码二维码

分享至好友和朋友圈