基于栅栏图像SVD相位相关算法的直线电机动子位置超精密测量研究

Linear motor servo system has broad application prospects in ultra-precision machining. The rotor position measurement precision of the linear motor directly determines the manufacturing precision of the system. This project proposed image measurement method to implement ultra-precision measurement of the linear motor rotor position by constructing fence image and utilizing the phase correlation algorithm with singular value decomposition（SVD）. Mainly researches are as follows: (1) the analysis of measurement error caused by linear motor rotor longitudinal vibrations, and then put forward the construction of fence image which is used for camera; (2) Based on gray gradient change of the fence image, Stripe width sequence of the fence image is obtained after improved Canny edge detection algorithm, taking the correlation coefficient and the variance of stripe width sequence as two eigenvalues, we select the fence image with stronger robustness as the target source image of high-speed camera; (3) we introduce the fence image phase correlation algorithm based on singular value decomposition, achieve sub-pixel position measurement of linear motor rotor, and propose the parallel processing method based on single-line calculation and multiple lines verification to greatly reduce the redundancy calculation and suppress noise. This measurement method has important theoretical and practical significance for the realization of the high-precision position feedback of linear servo system.

直线电机伺服系统在超精密加工中应用前景广阔。直线电机动子位置测量精度直接影响加工精度。本项目提出采用图像测量方法，通过构造栅栏图像和采用奇异值分解（SVD）的相位相关算法，实现对直线电机动子位置的超精密测量，提高测量的抗干扰性和实时性。主要研究内容如下：1）分析直线电机动子纵向振动对测量结果的影响，提出用于拍摄的栅栏图像结构；2）根据栅栏图像灰度梯度变化，以改进canny 边缘检测算法获得栅栏条纹的宽度序列，以宽度序列的“相关数组系数和标准差”为特征量，优选出鲁棒性较强的栅栏图像；3）引入基于奇异值分解的栅栏图像相位相关算法，实现对直线电机动子位置的亚像素测量，并提出基于多行抽样并行计算的校验算法，实现计算冗余度剔除和噪声抑制。该测量方法对于实现直线伺服系统高精度位置反馈具有重要的理论和实践意义。

直线电机伺服系统在超精密加工中应用前景广阔。直线电机动子位置测量精度直接影响加工精度。本项目提出采用图像测量方法，通过构造栅栏图像和采用奇异值分解（SVD）的相位相关算法，实现对直线电机动子位置的超精密测量，提高测量的抗干扰性和实时性。主要研究内容如下：1）分析直线电机动子纵向振动对测量结果的影响，提出用于拍摄的栅栏图像结构；2）根据栅栏图像灰度梯度变化，以改进canny 边缘检测算法获得栅栏条纹的宽度序列，以宽度序列的“相关数组系数和标准差”为特征量，优选出鲁棒性较强的栅栏图像；3）引入基于奇异值分解的栅栏图像相位相关算法，实现对直线电机动子位置的亚像素测量，并提出基于多行抽样并行计算的校验算法，实现计算冗余度剔除和噪声抑制。该测量方法对于实现直线伺服系统高精度位置反馈具有重要的理论和实践意义。