基于神经网络的组合导航系统状态估计

State Estimation of Integrated Navigation System Based on Neural Network

在干扰大的外界环境中,传统滤波法对组合导航系统进行状态估计的精度难以满足要求,为此提出了引入Elman神经网络。描述了它的状态估计的设计方法,对如何获取训练样本及网络的训练算法给予了详细的介绍,并把优化后的算法与原有方法进行仿真对比。最后以INS/GPS组合导航系统为例,分别用传统滤波法与Elman神经网络法进行状态估计。仿真结果证明了该法的有效性和实用性。

To estimate the state of integrated navigation system in serious interferential surroundings, the conventional filtering algorithms cant satisfy precision requirements. The paper introduces Elman network to resolve this problem and explains its designing means for estimating states. Its training algorithm is analyzed in detail and the approach to obtain swatch is also provided. Simulations are made by optimized algorithm and original algorithm. In the end, both the conventional filtering algorithm and the trained Elman network are used to estimate state of INS/GPS integrated navigation system. Simulation results show that the method is valid and practical.