Time grating sensor transforms space domain information to time domain and measures spatial displacement with time. To develop high precision time grating CNC rotary table and reduce the dynamic position feedback error of the table, circular position predictive measurement method is proposed for transforming time domain information back to the space domain based on time-space transformation technology. Predicted values are obtained by modeling the measured values with time series theory, and the last prediction error is corrected in real time using the current measured values. Modeling method and parameter estimation algorithm are presented. To confirm the validity of the position prediction method, an experimental system is designed. The dynamic prediction error of the rotary table circular position is 2, and precise predicting is achieved.