Research And Implementation Of Algorithm Of Heading And Attitude Reference System Based On MEMS

Attitude and heading reference system is an important branch of strapdown inertial navigation technology,which can provide real-time acceleration information,angular velocity information,magnetic vector information and attitude information for the further attitude control of the carrier.This thesis takes the attitude reference system based on lowprecision MEMS devices as the research object,to analyze its working principle focusing on how to improve its output accuracy in complex environments,so that to solve the problems of serious output misalignment or even invalidation of the system under strong interference.The thesis first introduces the output characteristics of the MEMS sensor.Then the thesis studies the working principle of the attitude and heading reference system,introduces the common attitude and heading calculation algorithms,and completes the design of the attitude and heading calculation part of this thesis.After the design of the basic part is completed,this thesis analyzes several interference environments often encountered by the AHRS in practical work,and understands their influence on the AHRS and their reflections in the sensor output data.Then an appropriate anti-interference scheme is designed according to the characteristics of interference.Finally,the function of the designed system is verified by the measured data.The main achievements of this paper are as follows:1.By analyzing the error source of the sensor during measurement,the measurement error model of the three-axis MEMS accelerometer and the three-axis MEMS gyroscope is established,and the measurement equation in matrix form and the sensor calibration scheme are given.2.The quaternion method is selected as the design method in the work,and the quaternion equation is analyzed by the fourth-order approximate Picard solution method to obtain the attitude change information.Then,this thesis determines the relationship between them and attitude based on the vector analysis of gravitational acceleration and geomagnetic intensity.Finally,a Kalman filter design based on quaternion is adopted to fuse the output of the accelerometer and the geomagnetometer with the updated attitude to obtain more accurate attitude information.3.Based on the classification function of BP neural network,an environmental disturbance detection scheme is designed,suitable data features are extracted from the outputs of accelerometer and geomagnetometer for the establishment of BP neural network model and subsequent applications.4.Aiming at the acceleration interference caused by the vibration of the carrier moving,the thesis adopts the smoothing filtering method.Since the frequency model of the vibration acceleration is not clear,the thesis designs an adaptive smoothing filter using a fuzzy controller to adjust the filtering step size.The unclear vibration acceleration is fuzzified by fuzzy theory,then,the filtering step size which is more consistent with the vibration acceleration characteristics is obtained by fuzzy reasoning.Because the characteristics of the vehicle’s maneuvering acceleration and magnetic field interference are similar,a unified treatment is made in this work.The residual between the sensor output value and the predicted value is used to dynamically adjust the weight of the acceleration and magnetic field strength measurement values in data fusion to reduce the misalignment measurement.The error caused by the value to the attitude solution.Based on the traditional combined vector self-test method,this work innovates and proposes an improved combined vector selftest method to detect the validity of data,which reduces the possibility of missing judgment by changing the steps of the original method.The test results show that the improved attitude and heading reference system has stronger anti-jamming ability compared with the original one,and the overall output error is reduced under the interference environment about 50%,and it can isolate the accelerometer and geomagnetometer information by itself when the interference is too large,so that the output error is kept within 5° in all process.The verification results are consistent with the expected design goals,it indicates that the system has high practical application potential.