Research On MEMS-SINS/GPS Integrated Navigation Based On Varitional Bayesian Filtering

In modern society,the integrated navigation system consisting of Strapdown Inertial Navigation System(SINS)and Global Position System(GPS)has been widely used,and with the development of Micro-Electro-Electro-Mechanical System(MEMS),MEMSSINS/GPS combined system has developed into an industry trend.Therefore,improving system accuracy in a variety of application environments has become a research hotspot.To slove this problem,this study mainly conducts research from the following two aspects:one is to study the compensation method of the conical error in the MEMS-SINS Euler angles calculation in a large dynamic environment;the other is to study the MEMSSINS/GPS tightly coupled nonlinear fusion method.Start with a single positioning source and improve the adaptability of the fusion filter to improve the positioning accuracy of the system.Based on the above research purposes,this study starts with the analysis of the classic multi-sample cone optimization algorithm in SINS attitude calculation,and proposes a correction algorithm based on the angular velocity rotation vector for the problem that it is not suitable for MEMS inertial sensors.The algorithm optimizes the coefficient of the angular velocity cross product.Improve the accuracy of attitude calculation.At the same time,a cone compensation algorithm that introduces the angular velocity correction of the previous period is proposed for it without considering the influence of the non-periodic component.The simulation test results show that under the premise of the same number of sub-samples,the two algorithms have absolute cone error compared to the classic error compensation algorithm.The value has dropped by 50%,and the accuracy of the two algorithms decreases as the cone half-angle and angular velocity of the cone movement increase,and increases as the attitude update cycle decreases.Aiming at the problem of nonlinear data fusion in the combined system,this study analyzes the characteristics of nonlinear filtering algorithms from two aspects: the expansion order of filtering accuracy and the calculation of floating-point number complexity,and finally adopts the high-precision and moderate-complexity volumetric Karl The Cabature Transformation Kalman filter(CKF)algorithm is used as the basic filter.To solve the problem of adaptive difference in the CKF algorithm,the Variational Bayes(VB)adaptive estimation method is introduced into the CKF algorithm to obtain the VB-CKF algorithm.Numerical simulation results show that the root-mean-square error of VB-CKF state estimation is reduced by twice that of CKF as a whole.The noise variance and steady-state performance are close to each other,but the convergence error of VB-CKF is 20% lower than that of CKF.The simulation results of the integrated navigation system show that VBCKF effectively suppresses the accuracy divergence of the CKF algorithm.The speed and position estimation error(RMSE)of the VB-CKF algorithm is reduced by 45%,and the attitude angle accuracy is improved by about 30%,but the time consumption of the VBCKF algorithm is increased by about 12%.Finally,this study builds a MEMS-SINS/GPS tightly coupled system platform on the existing platform,and solves the problem of MEMS inertial sensor and GPS clock synchronization by designing three clock interrupts.The actual measurement results show that when the variance of the GPS noise model is almost constant,the positioning performance of the VB-CKF algorithm is only improved by about 10% compared to the CKF algorithm,and the increase in accuracy and complexity brought about by it offset the increase.When the GPS noise model is unstable,the accuracy of the VB-CKF algorithm is improved by about 38% compared to the CKF algorithm.