Research On Error Compensation And Information Fusion Technology Of Micro-Inertial Sensor

With the development of micro-electro-mechanical system(MEMS),silicon based MEMS inertial measurement unit(MIMU)has made great progress,this kind of sensor has advantages of fast response,low power consumption and low cost,has been widely used in the field of navigation.Due to the temperature drift characteristics and processing technology of silicon sheet materials,MIMU precision is not high,and the error modeling compensation is the main method for improving the accuracy of MIMU.On the other hand,because of the MEMS inertial measurement unit integrated accelerometer and gyroscope,both error characteristics of each are not identical,using multi-sensor information fusion technology to become another effective way to improve the precision of MIMU.This paper studies the error compensation and information fusion technology of inertial measurement unit.First,on the basis of the working principle of the inertial measurement unit,this paper introduces the deterministic error model of inertial device,complete random error of the AR model and describes the modeling process;In the full study on the basis of basic principle of kalman filter,a concrete analysis of low dynamic environment state equation and measurement equation,puts forward a simplified model based on linear state equation of kalman filtering algorithm.In order to verify the error modeling experiment and data fusion algorithm of MEMS inertial measurement unit,the data acquisition system based on MPU9250 is also designed.The test results show that the random error mean square error is reduced from 0.07469°/s before compensation to 0.06781°/s after compensation,and the autocorrelation coefficient after compensation is reduced from 0.9982 to 0.0627,which effectively validates the random error compensation algorithm based on ARM A model.Correctness and effectiveness.In the static test and dynamic test,the heading angle error is within ±60,the pitch angle and roll angle error are controlled within±2°.The calculation time of one cycle is 0.0208s,which is only 58%of the complementary filter.The validity of the simplified Kalman filter is verified.With real time.