Adaptive Interacting Multiple Model Filter And Its Application On SINS Initial Alignment

The initial alignment technique is a main research point of the strapdown inertial navigation system(SINS),and the integrated alignment method which is based on the modern filtering theory has been widedly applied into the SINS alignment.In practice,the ship sometimes is required to finish the alignment process in sailing situation;however,the movements of the ship change severely which is affected by the ocean environment and it is hardly to meet the requirements on the alignment accuracy and time cost at the same time for the single mathematical model and filtering algorithm.Therefore,an adaptive interacting multiple model(IMM)algorithm focused on the linear and nonlinear models is proposed in the paper,in order to improve the SINS alignment performance when the ship is sailing in complicated ocean environments or the noise statistics of the filter are inaccurate.Firstly,the IMM theory is studied and the probability of the model interactions is derived in the paper.And then the IMM algorithm is tested in a target tracking simulation.Besides,the CDKF filter based on the second order Stirling interpolation is derived.To solve the estimation problems of the mixed linear and nonlinear system,the scheme that Kalman filter(KF)and CDKF are adopted respectively under the IMM framework is proposed,and a simulation test is performed in a linear/nonlinear system.Meanwhile,the performances of the IMM,KF and CDKF are analyzed.For details,both the mathematical model and the noise statistical knowledge are required to be accurately known.However,as the model uncertainties are contained in the system and the noise statistics are hardly to obtain in the practical applications,the performance of the conventional filters will degrade or even cause divergence.As a result,the innovation adaptive estimation Kalman filter(IAE-KF)and the innovation correction method based CDKF filter(ICM-CDKF)are derived respectively from the maximum likelihood estimation and innovation correction method in the paper,which is utilized to improve the performance of the IMM algorithm when the system contains model uncertainties,or the noise statistics are known or time varying.Moreover,the effects of the proposed algorithm are proved in a simulation test.In order to solve the sailing alignment problem in the tough sea environment or in the case that the noise statistics are inaccurate,the IMM adaptive filter is applied to the SINS alignment after the comprehensive considerations on the accuracy and time cost of the alignment process.Besides,the SINS error equation in motion base is derived,and thelinear/nonlinear state space equation of the small misalignment angle and large azimuth angle is obtained.In the SINS/GPS system,an improved ICM-CDKF algorithm is obtained under the linear measurement equation.Furthermore,to test the performance of the IMM adaptive filter,4 kinds of simulations that is the large azimuth angle,the small misalignment angle and the inaccurate noise statistical information in the sailing state and the large azimuth angle in the static base are implemented.The experimental results show that the IMM adaptive filter can shorten the alignment time,and do some help to improve the alignment accuracy.