制导工具误差分离自适应正则化模型及应用  

Adaptive Parameter Regularization Model for Guidance Instrumental Error Separation and its Application

作　　者：王文君[1] 段晓君[1] 朱炬波[1] 

机构地区：[1]国防科学技术大学理学院,长沙410073

出　　处：《弹道学报》2013年第3期28-33,共6页Journal of Ballistics

基　　金：新世纪优秀人才支持计划和航天飞行动力学技术重点实验室开放基金项目(SFDLXZ-2010-004)

摘　　要：在无地面测试信息条件下,针对制导工具误差分离的线性模型,提炼了基于稀疏约束的正则化模型一般形式。基于极大似然理论推导了自适应参数正则化模型,正则化参数可随逼近误差的变化而相应变化,避免了选取普适性正则化参数的困难。提供了基于拟牛顿方法的求解算法。实验结果表明,在同等条件下,该方法优于主成分分析方法和固定参数的正则化方法。Based on the linear guidance-instrument-error separation model,a general form of regu- larization model was carried out without the ground test information. A new adaptive parameter regularization model was deduced by maximal likelihood method. The regularization parameter is changed with the error of fidelity term, and it can avoid the difficulty of choosing a universal regu- larization parameter. An algorithm for regularization model was proposed based on Quasi-Newton method. The simulation results verifies that adaptive parameter regularization method is better than principal component analysis method and fixed parameter regularization method under the same simulation configuration.

关 键 词：制导工具误差 误差分离 正则化模型 稀疏约束 自适应参数 

分 类 号：V448[航空宇航科学与技术—飞行器设计]