FIR神经网络在导航数据滤波中的应用

FIR Neural Network’s application in data-filtering of navigation system

本文以寻北仪为例,介绍了一种新的惯性测量元件输出数据的滤波方案。即利用输出数据对一种叫做基于实时反向传播算法的FIR神经网络进行训练,继而通过神经网络来预测输出数据,再与实际输出数据加权平均,从而提高了数据的准确性。

This paper introduces a new Filter module for IMU＇s outputs in navigation system. The filter is based on FIR Neural Network using Temporal Backpropagation. Trained by IMU＇s history and current outputs, this network can make prediction about future ones, which improves Navigation computing＇s inputs when we compound the actual data with the prediction.