改进BP神经网络模式识别在消除惯导漂移误差中的应用  

Application in Eliminating Inertial Navigation Drifts Error Based on Pattern Recognition with Improved Back Propagation Neural Network

作　　者：武虎子[1] 南英[1] 武进雄[2] 吴欢欢[1] 

机构地区：[1]南昌航空大学航空与机械工程学院,江西南昌330063 [2]大连交通大学机械工程学院,辽宁大连116028

出　　处：《系统仿真技术》2008年第3期193-197,共5页System Simulation Technology

摘　　要：为了能实时地消除惯导陀螺漂移误差,在数字地图为基础的前提下,引入了改进BP神经网络模式识别技术,并设计了相应的三维数字地形辅助惯性组合导航算法——BP算法。在匹配过程中,分别给定了不同的数字地形模式和不同载体的机动轨迹。利用C++程序进行算法中的数值计算,然后通过Matlab软件模拟导航数值仿真。多次实验证明,提出的方法具有很好的可靠性和可行性,具有重要的工程应用价值和理论借鉴意义。In premise of basis of digital map, introducing the pattern recognition technology with improved BP neural network and designing aided inertial integrated navigation algorithm with corresponding three-dimension digital terrain-BP algorithm to enable eliminate INS （Inertial Navigation System） drifts error in real-time. Separately, different digital terrain pattern and different carrier maneuverable trajectory were given in process of matching. Numerical calculation was operated by C ＋＋ program and numerical simulation of navigation was done by Matlab software. According to several experiments ,the method introduced possesses good reliability, feasibility, important applicable value of engineer and reference significance of theory.

关 键 词：惯导漂移误差 数字地图 模式识别 BP神经网络 BP算法 数值仿真 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术] TP391.72[自动化与计算机技术—计算机科学与技术]