基于智能学习方法的弹载捷联惯导系统误差分析

Error Analysis of Missile-borne Strap-down Inertial Navigation System Based on Intelligent Learning Algorithm

针对弹载捷联惯性导航系统误差传播特性问题,提出基于智能学习的方法对系统姿态、速度和位置估计误差等进行分析。为抑制器件漂移和传统算法的高阶项近似误差等对系统精度的影响,在对捷联惯导系统器件误差模型、状态解算误差模型等分析的基础上,利用基于优化Elman神经网络的智能学习方法拟合系统非线性误差。在仿真试验中,基于设定的弹道,采集一定时间内系统量测和状态数据对模型参数进行迭代训练。最后,将训练好的网络模型用于系统状态估计,并与Kalman滤波方法进行比较,仿真结果表明所提方法具有更高的估计精度。

Aiming at the error propagation properties of missile-borne strap-down inertial navigation system,an intelligent learning algorithm is proposed to analyze the attitude,velocity and position estimation error of the system.In order to suppress the influence of device drift and high order approximation error of traditional algorithm on system accuracy,based on the analysis of device error model and state solution error model of SINS,the intelligent learning method based on optimized Elman neural network is applied to fit the nonlinear error of system.In the simulation experiment,on the basis of the set trajectory,the measurement and state data in certain time are collected to training the parameters of system model.At last,the trained network model are utilized to estimate the system state.The simulation results show that the proposed method have better accuracy in comparision of the Kalman filter.