A dynamic self-calibration method was proposed for an embedded time grating sensor to improve its measuring accuracy when it worked at changed electric parameters and measuring environments and could not be calibrated by a standard instrument. The features of the embedded time grating sensor were introduced, and a self-calibration method was presented by using error regular transformation of two discrete probes in a certain angle. Then, a dynamic self-calibration system was designed. To reduce the influence of sensor stability and environmental disturbance on the dynamaic calibrating processing, a Kalman filter was used. Moreover, an algorithm for residual control was proposed to ensure calibration precision. Finally, the self-calibration system was applied to calibration of the embedded time grating sensor.This calibration method was compared with the traditional standard method. Experiment results show that the original errors of sensor have been reduced from ±20″ to ±2.4″, which is nearly the same as the accuracy of standard instrument calibration and has been satisfied the calibration requirements of time grating sensors.