Initial Alignment Of SINS With Large Azimuth Misalignment Angles Based On Nonlinear Filtering

Strapdown inertial navigation system(SINS)is a dead reckoning navigation method.Initial alignment is one of the key technologies of SINS.The accuracy of the initial alignment directly affects the accuracy of navigation.A SINS is subject to random error of inertial sensor,system noise,modeling error,environmental disturbance,etc.If the small misalignment angle of SINS is still used,the accuracy of the estimated state will be affected.Therefore,for the large azimuth misalignment angle of SINS,a class of nonlinear filtering methods are proposed to the initial alignment.The main work and innovations of this thesis are as follows:For the poor estimation accuracy caused by the linearization of the extended Kalman filter(EKF),an iterative EKF-based initial alignment algorithm was proposed for SINS with large azimuth misalignment angles.The state estimation at time k is replaced by the counterpart at time k+1 in the proposed iterative EKF(IEKF)algorithm.The nonlinear equation is re-linearized in order to reduce the truncation error,which is caused by ignoring the high-order terms of the nonlinear function in EKF.For the case of large azimuth misalignment angles,the IEKF method can obtain better stability and higher alignment accuracy than EKF by adopting multi-step iterations on the state estimation value updated by the measurement.For the large azimuth misalignment angle the unknown system noise,an adaptive smoothing variable structure filtering(SVSF)approach is presented for SINS with large azimuth misalignment angle.The innovation adaptive estimation(IAE)is applied to estimate the variance matrix of measurement noise online.The current measurement data is fully used to identify the noise variance of the system in real time,reducing the calculation and shortening the alignment time.When the noise variance of SINS is unknown,the proposed adaptive SVSF can solve the parameter and model uncertainty problems better than EKF.Besides,the estimation accuracy of the azimuth misalignment angle is higher than that of EKF.For the unknown statistical characteristics of noise and the large azimuth misalignment angle,a particle filter(PF)combined with a SVSF is addressed for initial alignment of SINS.Since the SINS error model under large azimuth misalignment angle is no longer linear,and the dimension of state variables is large,the error model is necessary to be linearized by using SVSF alone.The larger the dimension of PF is,the lower its operation efficiency is.Therefore,the state vector of the system is divided into two parts.SVSF is used to filter the velocity and gyro errors,and PF is applied to filter the three attitude errors.This combined filter not only retains the advantages of PF,but also avoids the linearization process of SVSF.The proposed approach is more efficient than PF,and has better anti-disturbance ability than EKF in the case of non-Gaussian noise and abrupt-changed variance.The simulation results show that the proposed nonlinear filtering algorithms can improve the alignment accuracy and real-time performance for stationary base SINS with large azimuth misalignment angle.