Research On Random Error Modeling And Compensation Of MEMS Gyroscope

MEMS gyroscope has the advantages of miniaturization,light weight,low cost,high stability and high integration degree.MEMS gyroscope is limited by the existing device manufacturing technology,its measurement accuracy is relatively low and can only be applied to the low-end navigation field.The analysis of the error sources that affect the accuracy of MEMS gyroscope is an effective way to solve this problem.According to the random drift error is the main factor affecting the accuracy of MEMS gyroscope,this paper focuses on the research of the MEMS gyro random drift error compensation model.A modeling method based on time series Kalman filter was proposed,MEMS gyro random drift data are averaged,outlier,trend item extraction and parameter estimation,,then the Kalman filter compensation equation is established.Compared with the traditional method based on time series modeling,this paper proposed a new method of MEMS gyro random error compensation based on wavelet threshold de-noising and RBF network.The wavelet threshold denoising is used to separate the random drift of MEMS gyroscope,and the radial basis function neural network is used to predict and compensate the random drift of MEMS gyroscope.Allan variance analysis results show that the wavelet neural network modeling method outperforms the Kalman filter modeling method based on time series.