Modeling of MEMS Gyroscope’s Non-Stationary Random Drift

The random drift is the most important factor affects MEMS Gyro’s precision and it is anon-stationary random signal. For improving precision of gyro and reducing effect of random drift,using grey GM(1,1) model to extract established tendency signal, using power spectral density(PSD) to identify and extract the hidden periodic weak signal, then using AR(3) to model gyro’srandom drift. As the example of LCG50 MEMS gyro, the simulation experimental results show thatusing this method can reduce the random drift and enhance precision of gyro.