Research On Degradation Model And Prediction Method Of Acceleration Sensor

Quartz flexible accelerometer is the core components of inertial navigation systems. The reliability and stability of accelerometer directly affect the accuracy navigation of the overall system. The degradation model and Parameter Prediction of accelerometer have important value of application in the research of acceleration sensor's reliability and stability.Firstly, this paper analyzed several common degradation models of the acceleration sensor and designed an periodic test which combined centrifuge experiment and gravity eight angles flip experiment. Based on this experiment, using the existing experimental platform to collect Data has lasted for a year. The experimental data was filtered by kalman filter and preprocessed.Secondly, this paper, based on Auto-Regressive Integrated Moving Average Model(ARIMA), studied the prediction technique of the scale factor of accelerometer. Simulation results showed that the ARIMA model had low accuracy, high data fluctuation and conditional heteroscedasticity. In this paper, the Generalized Auto Regressive Conditional Heteroskedasticity model (GARCH) is used as the model of residuals, because of it's ability to processing data volatility and conditional heteroscedasticity of residuals. An improved ARIMA model which was embedded residual variance had been developed, which could solve the above problems, improve the modeling and prediction accuracy.Lastly, based on four kinds of evaluation criteria of prediction, the simulation results of the ARIMA and the proposed algorithm were comparative analysised. The simulation results showed that the proposed algorithm can effectively improve the accuracy of prediction parameters.