Research On Inertial/Geomagnetic Integrated Navigation Technology Based On NN-CKF

Geomagnetic navigation is a passive,autonomous,all-time,all-terrain navigation mode.Geomagnetic aided inertial navigation system can overcome the defect of the inertial navigation error which accumulated with the growing time,it also helps the system work autonomously in any environment and has a strong anti-interference ability,so the inertial / geomagnetic navigation system has broad application prospects in the military field.In this paper,we focus on the filtering technology of inertial / geomagnetic integrated navigation system.Firstly,this paper introduces the basic working principle of SINS and geomagnetic navigation system,and also introduces the mechanical arrangement and error equation of SINS,and then simulate the international geomagnetic reference model.Secondly,the inertial / geomagnetic integrated navigation system is modeled by the error equation of SINS and the international geomagnetic reference model.At the same time,according to the nonlinear characteristic of the measurement equation,several commonly used nonlinear Kalman filters(Extended Kalman filter,Unscented Kalman filter and Cubature Kalman filter)are briefly introduced,and through the one-dimensional non-stationary growth model and five target tracking model verifies that Cubature Kalman filter(CKF)has best filter performance.The feasibility of geomagnetic aided inertial navigation is simulated by CKF,and the simulation positioning accuracy is within a few hundred meters.Thirdly,aimed at the problem of inaccurate establishment of the geomagnetic field model due to the slow change of geomagnetic field,we introduces a neural network aided CKF algorithm.The idea of state expanding is used in this algorithm,and the weights of neural network are trained on-line by CKF,then the system model is modified by neural network.The simulation results show that the proposed algorithm has better performance than neural network aided EKF / UKF.Finally,the inertial / geomagnetic integrated navigation is simulated based on neural network aided CKF(NN-CKF).Firstly,considering the problem of inaccurate model,the feasibility of neural network aided CKF is verified by simulation,which can realize online correction of model error and improve the reliability of integrated navigation system.Then,according to the outlier problem,the system fault-tolerant is designed,through a neural network aided CKF fault-tolerant filtering algorithm which based on orthogonality of innovation,and the simulation results show that the fault-tolerant algorithm in the case of outliers can still output stability and improve the stability of integrated navigation system.