Research On Error Compensation And Data Fusion Of Mems Intertial Measurement Unit

Micro Electro Mechanical Systems(MEMS)is a new type of technology that integrates micro machinery,microelectronics,and micro nano processing technologies.The inertial measurement unit is a kind of micro inertial sensor.It uses micro mechanical technology to manufacture and integrate the gyroscope and accelerometer together,and can output the angular velocity and acceleration of the carrier and other information for automatic driving of cars,automatic tracking of robots and other aspects.In some extreme conditions,the output of inertial measurement unit will have large errors,and a single inertial measurement unit cannot guarantee the reliability of its output data when working in a complex measurement environment.Aiming at the above problems,this paper mainly studies the error compensation method of inertial measurement unit,and proposes an improved data fusion method based on D-S evidence theory,which is used to fuse multiple homogeneous inertial measurement unit to improve the stability and reliability of the system.The main research content is as follows:(1)Taking the MEMS inertial measurement unit RION16488 as the research object,this paper analyzes the working principles and performance indicators of the gyroscope and accelerometer in the inertial measurement unit,classifies the error types of the inertial measurement unit and expounds the error sources,which lays the foundation for the subsequent error compensation work.(2)Aiming at the problem of inertial measurement unit(IMU)deterministic error,a discrete calibration method is proposed.Firstly,the mathematical models of the gyroscope and accelerometer are established,and three main deterministic Error term are obtained.The angular rate rotation method of the gyroscope and the six position calibration method of the accelerometer in the discrete calibration are introduced in detail.Secondly,the STM32F103ZET6 processor and RION16488 are selected to form the master-slave structure.After establishing communication with the host computer through SPI protocol,calibration experiments are carried out on the rate turntable and the position turntable respectively,Finally,the effectiveness of the calibration algorithm was demonstrated by comparing the data effect and attitude calculation before and after calibration.(3)Aiming at the problem of random error of gyroscope in inertial measurement unit,an adaptive Kalman filter method is proposed to compensate for it.The improper selection of process noise matrix and measurement noise matrix in the traditional Kalman filtering method can lead to poor filtering effect.However,the adaptive Kalman filtering proposed in this paper can avoid the selection of two matrices and dynamically adjust the values of the two matrices according to the output values during the filtering process,making the compensation effect better.Experiments show that the adaptive Kalman filter has a significant effect on removing the five random errors of gyroscope:quantization noise,angle random walk,bias instability,angular velocity random walk,and rate slope.And compared with Kalman filtering,it was found that the compensation effect of this algorithm is significantly better than that of Kalman filtering.(4)Aiming at the problem of insufficient stability and accuracy of a single inertial measurement unit,an algorithm to improve the D-S evidence theory is proposed.This algorithm not only corrects the evidence,but also changes the combination method,solving the problem of evidence conflicts causing data fusion results to be incorrect in D-S evidence theory.And six sets of data fusion experiments were simulated under the static and moving states of gyroscopes,and the analysis of experimental results showed that the fusion effect was good.