基于灵敏度矩阵的高阶非球面自动干涉检测OA
Automatic null testing of high-order aspheric surfaces based on sensitivity matrix
高阶非球面高精度干涉检测是超精密光学加工的关键环节.传统手动干涉检测依赖专家经验,效率低且难以满足批量化检测需求.为此,基于刚体运动学与Fringe Zernike像差理论,推导了旋转对称面形的五自由度灵敏度函数严格表达式,对高阶非球面采用数值求积完成Zernike投影积分,避免了常规解析方法的低阶近似截断,仅需面形矢高参数即可完成建模,且无需经验标定或完整光学系统建模;针对传统最小二乘法无法解决面形误差与失调像差耦合的问题,提出一种加权最小二乘反演策略,通过增加权重因子对关键模态动态调整,有效抑制了面形误差对位姿反演的耦合污染,提高了检测精度.实验结果表明,已知失调量验证中五自由度校正误差不超过3.5%,未知失调量验证中传统方法无法正常收敛,光路失调引入的RMS误差为0.113λ,加权最小二乘经2次迭代后收敛,光路失调引入的RMS误差为0.004λ(λ=632.8 nm).该方法为高阶非球面批量化干涉检测提供了高效可靠的解决方案.
High-precision interferometric testing of high-order aspheric surfaces is a critical step in ultrapre-cision optical fabrication.Conventional manual interferometric testing relies heavily on expert experience,suffers from low efficiency,and fails to meet the requirements of batch inspection.To address these limita-tions,exact five-degree-of-freedom sensitivity functions for rotationally symmetric surfaces are derived based on rigid-body kinematics and Fringe Zernike aberration theory.For high-order aspherics,Zernike projection integrals are evaluated using numerical quadrature,thereby avoiding the low-order truncation in-herent in conventional analytical approaches.The proposed model requires only sag equation parameters as input and eliminates the need for empirical calibration or complete optical system modeling.To resolve the coupling between surface figure errors and misalignment-induced aberrations that cannot be effectively separated by standard least-squares methods,a weighted least-squares inversion strategy is introduced.In this approach,weight factors are applied to dynamically regulate the contributions of critical modes,there-by suppressing the influence of inherent figure errors on pose estimation and improving overall testing accu-racy.Experimental results demonstrate that,in known-misalignment verification,correction errors for all five degrees of freedom remain within 3.5%.In unknown-misalignment verification,the conventional method fails to converge,yielding a misalignment-induced residual of RMS=0.113λ,whereas the pro-posed weighted least-squares method converges within two iterations and reduces the residual to RMS=0.004λ(λ=632.8 nm).The proposed method provides an efficient and reliable solution for batch interfer-ometric testing of high-order aspheric surfaces.
郭乃泉;周满;张鑫;李明茁;胡海翔;薛栋林
中国科学院 长春光学精密机械与物理研究所,吉林 长春 130033||中国科学院大学,北京 100049||光学系统先进制造国家重点实验室,吉林 长春 130033中国科学院 长春光学精密机械与物理研究所,吉林 长春 130033||中国科学院大学,北京 100049||光学系统先进制造国家重点实验室,吉林 长春 130033中国科学院 长春光学精密机械与物理研究所,吉林 长春 130033||中国科学院大学,北京 100049||光学系统先进制造国家重点实验室,吉林 长春 130033中国科学院 长春光学精密机械与物理研究所,吉林 长春 130033||中国科学院大学,北京 100049||光学系统先进制造国家重点实验室,吉林 长春 130033中国科学院 长春光学精密机械与物理研究所,吉林 长春 130033||中国科学院大学,北京 100049||光学系统先进制造国家重点实验室,吉林 长春 130033中国科学院 长春光学精密机械与物理研究所,吉林 长春 130033||中国科学院大学,北京 100049||光学系统先进制造国家重点实验室,吉林 长春 130033
信息技术与安全科学
非球面检测灵敏度矩阵加权最小二乘自动对准
aspheric testingsensitivity matrixweighted least-squaresautomatic alignment
《光学精密工程》 2026 (9)
1400-1410,11
国家自然科学基金资助项目(No.62127901,No.62305333)
评论