基于时间序列的时栅数控转台动态建模研究

摘要

In order to apply a novel displacement sensor named time grating to full closed--loop CNC rotary table as an angle detector, it is necessary to transform temporal information to spatial information with time--space transformation algorithm for time grating sensor. According to time se- ries analysis theory, dynamic model of dynamic characteristics of time grating CNC rotary table can be established. Therefore, the next measurement value can be predicted based on past measurement values with the dynamic model. At a result, original absolute displacement signals sampled in equal time interval can be transformed to continuous incremental pulses required by full closed--loop CNC system. Furthermore, the principles of selection criteria for dynamic models, parameters estimation, model checking were presented. In addition, regarding error control, last predicted errors were always corrected by current measurement values in real time so as to eliminate accumulated errors for realizing precision measurement. Experimental results prove that dynamic measurement displacement errors are within ±3＂ with the dynamic model. And precise full closed--loop angular displacement measurement is realized.%时栅运用到全闭环数控转台作位置检测传感器时，需采用时空变换算法将时栅的时域信息转换成空域信息。利用时间序列对时栅数控转台的动态特性进行建模，依据一系列过去测量值预测下一采样时刻角位移，将原本等时采样的绝对式角位移转换为全闭合数控系统需求的等空间增量式连续脉冲。介绍了动态模型选择标准、参数估计、模型检验的原理和算法。采用当前预测值对上一次预测误差进行实时修正，以消除累计误差保证测量精度。实验证明了采用动态模型预测算法能保证动态数控角位移测量误差控制在±3＂以内，实现了精密动态全闭环角位移测量。