Modeling Of Micromachined Silicon Gyro Random Drift

With the improvement of the inertial system precision. the research on gyroscopic drift has been more and more important. As a neotype gyro, microma-chined silicon gyro has many advantages, such as small size, light weight, low cost, low power consumption, long life period, high reliability and good dynamic performance etc, which are much more superior than traditional inertial device. However, it has a large random drift at the present time, and its accuracy can not be greatly improved in a short time. Given these conditions, its random drift can be tested, analyzed and modeled based on time series. At the basis of this model, random drift can be forecasted and compensated so that the micromachined silicon gyroscopic precision can be improved.Focusing on micromachined silicon gyro random drift modeling, the main and creative achievements of this paper include the following aspects:At first, the structure of time series model and their character are introduced. Then, based on the test data, statistical character including stationary, normality, tendency and periodicity is systematically analyzed in detail.Methods for AR model and ARMA model parameter estimation were discused in detail. compared with several recursive algorithms, nonlinear least square method had been selected to estimat the parameters.Based on the statistical character of random drift, random drift had been modeled using conventional method and DDS method respectively. Suitability test had been also made for the model.Finally, based on the model, minimum variance forcast method has been used for random drift forecasting.