Abstract: According to the characteristics of traditional systematic calibration method that of high state variable dimension, poor observability and easy filtering divergence, a two-level systematic calibration method was proposed. By merging the accelerometer calibration errors and FOG calibration errors, the dimension of Kalman filter was reduced. The systematic calibration method took speed error and attitude error as observation vector and designed six positions to estimate the state vector, then decoupled the estimation and calculated the calibration errors. For the characteristics that the estimated time of FOG calibration error was long and the beginning of Kalman filter convergenced rapid, it estimated the FOG calibration error twice and took the sum of twice estimations as FOG calibration error. The simulation and experiment results show that two-level systematic calibration method can calibrate twenty-four errors of calibration parameters including scale factor, installation error and zero and improve position accuracy of system.

Key words: two-level systematic calibration method, scale factor, installation error, zero