| Inertial navigation system(INS)is a set of calculated navigation system,using gyroscope and accelerometer as the inertial sensitive devices.Integrating the angular velocity and specific force measured by sensitive devices,the system's current attitude,velocity and position information can be obtained.The performance of INS depends largely on the accuracy of the initial alignment.Transfer alignment(TA)is an important method of fast initial alignment.This process is generally accomplished by calculating the difference between a slave inertial navigation system(SINS)and a master inertial navigation system(MINS)to form observations which are then used in a Kalman Filter(KF)to recursively estimate and compensate SINS attitude and inertial sensor errors.In this paper,the performance optimization of the Strapdown INS(SINS)is carried out in terms of the TA error model,matching mode,filtering algorithm and external information auxiliary alignment.First of all,the paper provides and analyzes the key technologies involving TA.An integrated error model is established considering dynamic deflection and lever arm effect at the same time.Based on this,the velocity error equation and attitude error equation in the existing literature are rederived and perfected.Secondly,three algorithms of different measurement parameters matching and calculation parameters matching for TA scheme are designed in detail,and theoretical analysis and simulation verification are carried out respectively under low dynamic conditions.On that basis,an improved method of integrated velocity and integrated angular velocity matching is proposed to further improve the accuracy and rapidity of TA.The advantages and disadvantages of various algorithms are summarized through the comparison and analysis of the estimation error of fixed mounting angle and deflection angle.Thirdly,for the problem that in the process of the actual TA,the system dynamic model is not accurate and the noise statistical characteristics are difficult to accurately determined,the existing Sage-Husa adaptive Kalman Filter(AKF)and the maximum likelihood AKF are improved separately.The system noise and measurement noise real-time estimators are constructed so that the model parameters and noise statistics can be updated and corrected with observation data.At the same time,an UD decomposition algorithm is introduced and used in one-step prediction error covariance matrix,and the decomposition matrix is applied to the updating process of the measurement information to ensure its nonnegative definiteness,and thus further enhancing the robustness of the filter.Finally,in the case of the low accuracy or malfunction of MINS in the TA process,it is necessary to study the auxiliary TA of external observation information in order to ensure that the mission can be completed reliably.An Star/INS matching TA scheme is adopted.The inertial information is fully calibrated using the high-precision star azimuth data information and the feasibility of this method is verified by simulation. |
