基于双远心立体视觉DIC方法的板料成形局部高梯度应变精确测量研究

The marcro global material flow and local deformation concentration are two critical process in sheet metal forming. The accurate measurement of the local high gradient discontinuous deformation has grate importance on characterizing the local strain evolution law and building fracture criterion. The traditional 3D DIC focuses on the global strain measurement, it can hardly tackle the complex local strain due to the poor local space resolution and imperfect subset tracking ability, but few related research is reported on this problem. Based on the single-lens stereovision and bi-telecentric lens, a novel bi-telecentric stereovison DIC method is proposed. We will firstly study the imaging principle of the proposed stereovision, and try to build the accurate projection model based on telecentricity, and find the key paramenters on the accuracy of reconstruction. Then, the bi-prism distortion model will be established, and the accurate and reliable stere-calibration algorithm will be further studied. Meanwhile, the high order subset for high gradient deformation will be proposed as well as the splitting strategy for discontinuous deformation, and invere compositional algorithm for this type of subset will also be derived. Finally, a local complex strain filed measurement system will be designed for sheet metal forming, and the evolution progress of typical local complex deformations will be investigated, the measurement accuracy will be verified. The proposed method will become a power member in the measurement technique family in sheet metal forming field.

金属板料成形是材料总体宏观流动和局部集中变形的有机结合和相互协调的过程。局部高梯度、不连续变形的高精度测量对于描述局部应变演化规律、建立断裂准则具有重要意义。传统三维DIC方法主要关注全局变形测量，因受限于局部空间分辨率和子区跟踪能力，其处理板料局部复杂变形能力不足，而该方面的目前研究十分缺乏。为此创新性提出了基于单镜头立体视觉和双远心镜头的双远心立体视觉DIC方法。本项目拟首先研究双远心立体视觉系统的成像机理，建立基于远心度的高精度双远心投影模型，揭示影响三维重构精度的本质原因，其次从畸变机理角度提出参数化的双棱镜畸变模型，研究高可靠性的成像系统标定模型及算法，进一步构建针对高梯度变形的高阶子区，提出处理不连续变形的裂解机制，推导高阶形函数的逆复合迭代公式，最后开发针对板料局部复杂应变测量的DIC技术，研究典型局部复杂变形的演化规律并验证测量精度，完善和丰富现有板料成形研究的测量手段。

研究金属板料成形的过程中，局部高梯度、不连续变形的高精度测量对于描述局部应变演化规律、建立断裂准则具有重要意义。传统三维DIC方法受限于局部空间分辨率和子区跟踪能力，导致其三维局部复杂变形能力不足。为此创新性提出基于双远心镜头和双棱镜的双远心立体视觉DIC方法。本课题从成像及标定模型、高梯度应变平滑算法、并行求解算法和裂尖变形场测量方法四方面展开了深入研究。1）在成像及标定模型方面，根据双远心成像模型推导出了双远心标定系统算法，考虑了双远心镜头的畸变模型，对于双棱镜的成像模型，创新性地利用Snell公式推导出空间中任意位置入射和二次折射光的几何关系，提出双棱镜的姿态参数这一概念，推演出双棱镜-双远心立体成像系统的模型；2）在高梯度应变平滑算法方面，使用Hermite单元形函数和高阶多项式作为核函数，先后提出LH算法和RPS算法，两种方法均利用Tiknonov正则化和GCV函数减少噪声的影响和降低对窗口的敏感度，其中RPS方法简单易操作，精度高于传统方法，且不易受到参数选择的影响，特别适合于解决金属板料成型中高梯度复杂应变场测量问题；3）在并行求解算法方面，在路径依赖DIC算法的框架下，提出基于多核CPU的高性能并行DIC测量算法，该算法成功实现了单相机5000点以下计算区域的实时测量，速度比单线程方法高一个量级；4）对于裂尖变形场测量，结合DIC提出了一种基于变形梯度的快速确定裂纹尖端位置的方法，同时还采用高阶Hermite形函数构建全局DIC的有限元单元，建立同时考虑灰度误差与位移场光顺性的HRGDIC算法，大大提高了不连续高梯度变形场计算精度。综合上述理论研究成果，进一步搭建了一套双远心立体视觉DIC测量系统，可有效解决板料成形中局部高梯度和不连续应变场测量问题，进而实现了本课题的研究目标。