In this paper, we propose an intelligent segment-fitting algorithm to increase the measurement accuracy of quadrant detector (QD) in straightness measuring system. Add-subtract algorithm and polynomial fitting are usually adopted to acquire the estimate of the spot projected on QD. The accuracy of QD is limited by the nonlinear relationship between the spot position and its estimate. In order to obtain higher accuracy, the polynomial fitting order must be higher with more calculation amounts, so the new algorithm is designed to optimize the nonlinearity. Through the specific method mentioned in the new algorithm, the whole measurement range of QD can be divided into several segments intelligently. And the data in each segment and between adjacent segments can be simply expressed by the third order polynomial fitting and linear interpolation. The experiment results verified that the algorithm could increase the measurement accuracy with good real-time performance. Within the measurement range of ±250 μm, the residual RMSE of the QD using the proposed algorithm, compared with that using the third order polynomial fitting, was reduced more than 80%, and the calculation amount was reduced to about 1/4 of that using the seventh order polynomial fitting.

本文提出了一种智能的段拟合算法，以提高直线度测量系统中象限检测器（QD）的测量精度。通常采用加减法和多项式拟合来获得对QD投影点的估计。QD的精度受光点位置与其估计值之间的非线性关系限制。为了获得更高的精度，多项式拟合阶数必须更高且计算量更多，因此设计了新算法来优化非线性。通过新算法中提到的特定方法，可以将QD的整个测量范围智能地分为几个部分。并且通过三阶多项式拟合和线性插值可以简单地表示每个段中以及相邻段之间的数据。实验结果证明，该算法可以提高测量精度，实时性好。在±250μm的测量范围内，与使用三阶多项式拟合的方法相比，使用所提出算法的QD的残留RMSE减少了80％以上，并且计算量减少了约1/4使用七阶多项式拟合。