Application of the Digital Signal Procession in the MEMS Gyroscope De-drift

The Micro Electrics Mechanics System (MEMS) gyroscope has the virtue of lower price, smaller size, lighter weight than of the traditional rate sensors. It can be widely used in the low-price inertial navigation system. But limited by its machining precision, the MEMS gyroscope usually has the large drift. Using the digital signal procession (DSP) of drift signal, the performance of the MEMS gyroscope could be improved effectively. By a new median filtering method, the outlier signal from the drift signal was got rid of. The online implemented wavelet analyzing of the drift signal was used to extract the character of the gyroscope drift. A two-order AR model was built for the residual signal between the original signal and the characteristics drift signal. A group of matrix equations were built for the traditional Kalman filter. By making some improvement on these equations, an adaptive Kalman filter was used to trace the change of the residual signal adaptively and could increase the de-drift effect of the MEMS gyroscope greatly. A 50mm times 50mm circuit board was made to implement the DSP. Standard variance of the dynamics signal measurement was 18.31 deg/h after the gyroscope drift was extracted