利用神经网络和卡尔曼滤波的组合导航技术  

The Technology of Integrated Navigation Based on ANN and Kalman Filter

作　　者：屈重君[1] 郭志想[1] 陆志东[1] 

机构地区：[1]西安飞行自动控制研究所,西安710065

出　　处：《传感技术学报》2007年第11期2504-2507,共4页Chinese Journal of Sensors and Actuators

基　　金：航空基金资助项目(03I18007)

摘　　要：Kalman滤波是组合导航中最常用的最优滤波工具,但在组合导航系统中仍有一些应用的局限性.针对卡尔曼滤波受到需要已知精确系统模型的限制,文中利用神经网络在解决非线性、时变系统、不确知系统中的独特优势,提出一种新的Adaline/Kalman组合滤波方法,阐述了神经网络的选取,网络的训练,给出具体仿真条件,应用于卫星/惯性组合导航系统.文末给出了仿真结果,表明滤波的效果较之单独的Kalman有所改善,导航精度有所提高,同时也说明了提出的整个方法是可行的,正确的.Although the Kalman filter was the optimum filter used in integrated navigation system,it also have some application limitations.Artificial Neural Networks(ANN)has been widely used in various controlling systems after its naissance.Nonce it has been a quite completely science in control field and solved a lot of different controlling problems in nonlinear system and uncertainly system,especially in parameter estimate,state estimate,alignment adjustment,integrated navigation filed etc.A new integrated filtering method based on Adaline /Klaman is presented that takes advantage of both Adaline and Kalman filter,and simulation condition,networks choice and practice is given.A better and higher navigation accuracy was gained as we expected,the simulation result shows that the new method is reliable and correct.

关 键 词：神经网络 组合导航系统 卡尔曼滤波 自适应神经网络 

分 类 号：V249.1[航空宇航科学与技术—飞行器设计]