Study of the GM(1,1) Model and Improved Model of the Trend Term of Micromechanical Gyroscope Drift

To improve measure precise of micro inertial integrated navigation system, modified GM(1,1) using exponential smooth is proposed to depict nonstationary random drift signal of micromechanical gyroscope under zero input condition. The model is compared with ones build using linear regression and polynomial regression methods. The experiment and respected simulation results show that the modified GM(1,1) model is better for eliminating the trend term of nonstationary signal and the extracted signal is proved to be zero-mean, normal and stationary random series.