Research On Application Of Particle Filtering Algorithm To Sins Initial Alignment

Initial alignment is an important technology in inertial navigation, the accuracy and the using time of the alignment play an import role in the working performance of the inertial systems. At present, Kalman filtering is the most widely used method in the Strap-down Inertial Navigation Systems (SINS), but Kalman filtering need to linearize the nonlinear system model and it demands that the noisy be Gaussian or approximate to Gaussian. All of the factors above restrict the application in practice but also reduce the accuracy of the filtering. Particle filtering can solve nonlinear/non-Gaussian filtering problem, and for the reason, particle filtering and its application in SINS initial alignment were discussed.To investigate on the initial alignment of the SINS system, firstly, the principle and the key steps of the particle filtering were investigated deeply, and the problem of degeneracy and sample impoverishment were proposed.Then, to improve the performance of particle filtering, the algorithm was improved by two different strategies. To get better importance density for variant systems, the tradition EKF algorithm was introduced and got the EKPF. On the basis of EKPF, to get better filter performance in the environment of model uncertainty, the EHF was combined with particle filtering and the EHPF was got. The degeneracy and sample impoverishment of the particles is also an important problem to be solved, by introducing the genetic operation of Genetic Algorithm into particle filtering, the GPF algorithm was obtained.At last, the simulation of the particle filtering and its application in initial alignment of SINS system were performed. The simulation results show that particle filtering is more accurate than EKF in the alignment, and the improved algorithms can get better alignment accuracy.