Research On Multi-sensor Data Fusion Algorithm For AUV Attitude Detection

Attitude detection is the basis of navigation and control of AUV.It is very important to accurately measure the attitude information of AUV and make the corresponding control information to ensure the basic navigation of AUV.In view of the limitations of the existing multisensor data fusion algorithm,this paper focuses on the unscented Kalman filter algorithm for nonlinear systems,proposes a multisensor data fusion algorithm based on AUV attitude detection,adjusts the noise covariance through adaptive strategies,improves the accuracy of the fusion algorithm,and verifies its performance through experiments.Firstly,this paper analyzes the development of data fusion algorithm,and compares and summarizes the attitude algorithm and fusion algorithm.Secondly,the basic theory of AUV attitude detection is studied from the selection of coordinate system and the representation method of attitude,focusing on the implementation process of attitude update based on quaternion.By studying the principle of MEMS sensor attitude measurement,the method of attitude measurement based on sensor combination is designed.In the phase of AUV attitude data collection,the characteristics of AUV system and the performance of UKF Algorithm in strong nonlinear environment are analyzed firstly,and the data fusion algorithm in this paper is determined.A small AUV attitude measurement system is designed by using low-cost MEMS sensor,and reliable original attitude data is obtained by sensor configuration.By analyzing the error sources of the sensor,the error model is established,and the preprocessing of the original attitude data is realized to improve the measurement accuracy.In the research and design stage of data fusion algorithm,the AUV attitude data fusion algorithm proposed in this paper mainly focuses on the optimization and design of UKF algorithm from two aspects:system model simplification and adaptive factor research.The quaternion updating equation is used to replace the nonlinear state equation in UKF algorithm to reduce the number of UT transformations.The adaptive factor is set by comparing the size of the theoretical covariance matrix and the actual covariance matrix of the residual,and then the system noise covariance and the measurement noise covariance are adjusted by using this factor to solve the error impact caused by the uncertainty of noise statistical characteristics.Finally,see details The implementation steps of UKF algorithm after the fusion optimization strategy are introduced in detail to achieve the fusion processing of sensor attitude data and the improvement of accuracy.Finally,the attitude data fusion algorithm is simulated and verified.The experiment is divided into static test and dynamic test.In static test,the fusion accuracy under fixed attitude angle is analyzed.In dynamic test,dynamic data is collected for fusion analysis.The experimental results show that the data fusion algorithm based on UKF designed in this paper can avoid the gyro drift error and improve the fusion accuracy effectively.And the attitude determination system system has the characteristics of miniaturization and low power consumption,which can meet the needs of civil AUV and has certain application value.