|Table of Contents|

Design of Adaptive Federated H Filter for Rapid Transfer Alignment


Research Field:
Publishing date:


Design of Adaptive Federated H Filter for Rapid Transfer Alignment
HU Jian1MA Da-wei1CHENG Xiang-hong2ZHOU Bai-ling2
1.School of Mechanical Engineering,NUST,Nanjing 210094,China;2.School of Instrument Science and Engineering,Southeast University,Nanjing 210096,China
rapid transfer alignment federated H∞ filter adaptive algorithms Elman network
Considering the high filter order,the huge calculation burden,the poor robustness and the low accuracy of the centralized Kalman filter for rapid transfer alignment,a federated H∞ filter is presented here to realize rapid transfer alignment.The structure and algorithm of the federated H∞filter is designed.The filter employs a two-stage data fusing architecture which contains two sensor-dedicated local filters and a master filter to fuse the local filters’ outputs.The whole task is assigned to be performed synchronously by two local filters.The calculation burden is alleviated.An improved Elman network is proposed to adjust information sharing coefficients adaptively and detect subsystem malfunction to realize the adaptive share of fusing information in each subsystem and isolate the malfunctioning subsystem that has been detected.Simulation results show that the calculation speed and robustness is improved and the alignment accuracy is about one order of magnitude higher by using this method.


[1] Shorte lle K J, Graham W R, Rabourn C. F??16 flight tests o f a rapid transfer a lignment procedure[ A ]. IEEE Pos ition Loca tion and Nav igation Symposium [ C ]. Pis?? ca taw ay, N J, USA: IEEE, 1998: 379- 386.
[2] Lyou J, L im Y C. Transfer alignm en t erro r com pensa?? tor des ign based on robust state estim ation[ A] . Trans?? actions of the Japan Soc iety forAeronautica l and Space Sc iences[ C]. Tokyo, Japan: So ciety for Aeronautica l and Space Sciences, 2005: 143- 151.
[3] Gu D Q, Q in Y Y, Peng R, et a.l Rapid transfer a?? lignm ent us ing fede rated Ka lm an filter [ J]. Transac?? tions o f Nanjing Un ive rsity o f A eronautics and A stro?? nautics, 2005, 22 ( 2) : 139- 143.
[4] Yan J G, Yuan D L, W ang XM, et a.l A study o f in?? forma tion fus ion based on federa ted K alm an filter ing [ A]. Pro ceedings o f theW orld Cong ress on Intelligent Contro l and Autom a tion [ C ]. Piscataway, NJ, USA: IEEE, 2006: 1474- 1477.
[5] Jam sha id A, Fang J C. SINS /ANS /GPS integration u?? sing federated K alm an filter based on optim ized in fo r?? m a tion??shar ing coe fficients[ A] . AIAA Guidance, Nav?? iga tion, and Con tro l Con f[ C ]. Reston, VA, USA: AIAA, 2005: 6028- 6040.
[6] 单茂华, 周百令, 黄丽斌. 联合H?? 滤波算法及其在 组合导航系统中的应用研究[ J] . 数据采集与处 理, 2003, 18( 4): 418- 422.
[7] 程向红, 万德钧. 分布式系统中捷联惯性系统动基 座对准研究[ J]. 中国惯性技术学报, 2004, 12( 6): 8- 12.
[8] 谷吉海, 姜兴渭, 王晓锋, 等. 双向E lm an神经网络 在卫星电池阵功率预测中的应用研究[ J]. 南京理 工大学学报(自然科学版), 2004, 28( 4) : 404- 408.
[9] H eH T, T ian X. An improved E lman ne tw ork and its application in fla tness pred iction m ode ling[ A]. Second Inte rna tiona l Con ference on Innovative Computing, In?? form ation and Contro ,l ICIC IC 2007 [ C ]. P isca taw ay, N J, USA: IEEE Compute r Soc iety, 2008: 44- 48.
[10] Cheng Y C, Q iW M, Ca iW Y. Dynam ic properties of E lm an and m odified E lm an neura l netwo rk [ A ]. Pro ceedings o f 2002 Internationa l Conference on M a?? ch ine Learn ing and Cybernetics[ C]. P iscataw ay, N J, USA: IEEE, 2002: 637- 640.
[11] Moham ad F A, H aza N A H, S itiM H S. Augmentation of E lman recurrent network learning w ith partic le swarm optmi ization [ A]. Second Asia Internationa l Conference onMode ling and S mi ulation[ C]. Piscataw ay, NJ, USA: IEEE Computer Society, 2008: 625- 630.


Last Update: 2010-06-30