文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > 自然科学 > 数学 > 有色噪声干扰随机系统的偏差补偿递推最小二乘辨识方法
侵权投诉

有色噪声干扰随机系统的偏差补偿递推最小二乘辨识方法

For ARX-like systems, this paper derives a bias compensation based recursive least squares identification algorithm by means of the prefilter idea and bias compensation principle. The proposed algorithm can give the unbiased estimates of the system model parameters in the presence of colored noises, and can be on-line implemented.

有色噪声干扰随机系统的偏差补偿递推最小二乘辨识方法

YongZhang ,GuimeiCui

SchoolofInformationEngineering,InnerMongoliaUniversityofScienceandTechnology,Baotou014010,PRChina

articleinfoabstract

ForARX-likesystems,thispaperderivesabiascompensationbasedrecursiveleastsquaresidenti cationalgorithmbymeansofthepre lterideaandbiascompensationprinciple.Theproposedalgorithmcangivetheunbiasedestimatesofthesystemmodelparametersinthepresenceofcolorednoises,andcanbeon-lineimplemented.Finally,theadvantagesoftheproposedbiascompensationrecursiveleastsquaresalgorithmareshownbysimu-lationtests.

CrownCopyrightÓ2010PublishedbyElsevierInc.Allrightsreserved.

Articlehistory:

Received26November2009

Receivedinrevisedform21September2010

Accepted4October2010

Availableonline16October2010Keywords:

ARX-likesystems

ParameterestimationRecursiveidenti cationLeastsquares

Biascompensationprinciple

1.Introduction

Basedonobserveddatafromthesystems,systemidenti cationdealswiththeproblemofbuildingmathematicalmodelsofdynamicalsystems[1],inotherwords,identi cationistoestimatetheunknownparametersofsystemsbyusingthemea-suredinput–outputdata.Theleastsquares(LS)methodhasbeenthedominantalgorithmforparameterestimationduetoitssimplicityinconceptandconvenienceinimplementation[2].However,ifthedisturbancesactingontheplantarecolorednoises,thentheLSmethodisunabletogiverisetoconsistentestimatesofthesystemparameters.Overthelasttwodecadesmuchefforthasbeendevotedtotackletheproblemoftheestimationbiasarisingfromcorrelateddisturbances[3],andtheresearchhasledtothedevelopmentoftheinstrumentalvariable(IV)methodsandthepredictionerror(PE)methods,namely,LSbasedmethods,forexample,recursiveextendedleastsquares(RELS)method,recursivegeneralextendedleastsquares(RGELS)method[4],stochasticgradient(SG)algorithm[5–7]etc.TheIVmethodsarefavorablyadoptedinthepres-enceofinaccuratenoisemodelswithoutrequiringtheinformationofthenoisecharacteristics.Butthechoiceofinstrumentvariablesmayaffectultimateidenti cationresultssubstantiallyincertainsituations,andtherehasnotbeenanyef cienttechniqueof ndingoptimalinstrumentalvariables.TheRGELSmethodisessentiallyaprocedureforsolvingahighlynon-linearminimizationproblem.Itmaynotconvergetotheoptimalsolutioninsomecases,especiallyifthesignal-to-noiseratioistoolowortoohigh.TheconvergenceoftheRELSmethodaswellasthePEmethodwilllargelydependonwhetherornotnoisemodelssatisfycertainstrictlypositiverealcondition.TheSGalgorithmrequiressmallcomputationaleffort,buthasslowerconvergencespeedcomparedwiththeleastsquares.

Apartfromtheaforementionedtwoprimarykindsofunbiasedestimationapproaches,i.e.,thePEmethodsandtheIVmethods,thereareothertypesofiterativemethods[8–10],multi-innovationmethods[11,12]andbiascompensationbasedmethods[13–16]thathavereceivedconsiderableattention.Theiterativemethodshavegoodperformanceforlinearor

Correspondingauthor.

E-mailaddress:yinghuzhangyong@163.com(Y.Zhang).

0307-904X/$-seefrontmatterCrownCopyrightÓ2010PublishedbyElsevierInc.Allrightsreserved.doi:10.1016/j.apm.2010.10.003

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email