基于散斑图像MLS融合算法的直线电机动子位置精密测量研究

Real-time precision measurement of the position of the linear motor mover is the key to realize the precise positioning of the linear servo control system. This project proposed digital speckle correlation measurement method to implement precision measurement of the linear motor mover position by using non-parametric modeling to the optimized speckle image and introducing Kalman filter to improve the curved surface fitting algorithm. The main researches are as follows：(1) The effects of the speckle image parameters such as gray-scale gradient, speckle size and number of speckle images on the measurement accuracy and measurement efficiency are studied. The standards of speckle image quality evaluation based on Pantt entropy and Peak Signal to Noise Ratio are also established. (2) According to the above parameters and standards, a non-parametric modeling (k-Nearest Neighbor method) is carried out to construct speckle images which can satisfy the robustness requirement and reduce the calculation redundancy. (3) A sub-pixel measurement algorithm based on Kalman filter and moving least squares curved surface fitting is proposed, and the precision measurement of the mover position is realized. This measurement method has important theoretical and practical significance for the realization of the high-precision position feedback of linear servo system.

直线电机动子位置的实时精密测量是实现直线伺服系统精确定位的关键。本项目采用数字散斑相关测量技术，通过非参数建模优选散斑图像和引入Kalman滤波改进曲面拟合相关算法，实现对直线电机动子位置的精密测量，并提高测量的稳定性和实时性。主要研究内容如下：1）分析散斑图像的特征参数对直线电机动子位置测量精度和实时性的影响，提出采用Pantt熵和PSNR作为散斑图像质量的评价标准。2）根据散斑特征参数和评价标准进行k-最近邻(kNN)算法建模，构造出既能满足抗干扰性要求又能降低计算冗余度的散斑图像。3）提出将Kalman滤波融合至移动最小二乘（Moving Least Square, MLS）曲面拟合亚像素测量算法中，结合系统标定实现对动子位置的精密测量。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。

直线电机动子位置的实时精密测量是实现直线伺服系统精确定位的关键。本项目采用数字散斑相关测量技术，利用灰度梯度和优选散斑图像和引入Kalman滤波改进曲面拟合相关算法，实现对直线电机动子位置的精密测量，并提高测量的稳定性和实时性。取得的主要进展及重要结果如下：1）设计并搭建了直线电机动子位置图像测量系统，安装高精度光栅测量系统作为校验信号，用于后续图像检测结果与光栅测量结果进行对比，从而验证测量方法的有效性。2）研究散斑图像的灰度分布、散斑尺寸、数目、空间分布参数对测量精度和实时性的影响。通过理论推导，提出了以散斑灰度梯度和作为散斑质量评价指标，用以筛选目标图像，提高测量鲁棒性。3）考虑直线电机运行过程中的纵向振动所引起的扫描偏差，研究了目标图像的抗振性能。将筛选后的散斑图像优化为栅栏图像，进一步提高了检测的稳定性。4）研究曲面拟合亚像素测量算法及其抗噪性能，提出一种联合相位差算法，采用空间域灰度匹配获得整像素位移，再利用频域相位差获得亚像素位移，实现动子位置精密测量；研究曲面拟合算法的测量量程，提出了移动最小二乘（Moving Least Square, MLS）时间图像相关亚像素测量算法，满足直线电机在高速运行时的测量精度的稳定性。结果表明，所提方法可以在不同运动模式下精确跟踪直线电机动子位置，平均测量精度达到0.006mm。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。