Abstract: When mobile special maneuver platform working on the long-term tracking,fiber-optic gyroscope can meet the demand in the space due to its advantages.Fiber-optic gyroscope self calibration should be done in the space.High fidelity gyro models cost more computation time and usually reduced module filter lost navigation precision.Aimed at the problem,using sensor observation to analyze the high-dimension gyroscope module and taking the states with low observability as model errors,a reduced-dimension module based on module prediction was proposed.The proposed module could guarantee the precision,decrease the computation and improve the algorithm speed.The hardware-in-the-loop simulation results show that the algorithm has higher precision than that of the high-dimension self calibration algorithm,and its computational cost is only 1/10 of the full-dimension model.The algorithm also improves the real-time performance.