question: kalman filter

Article View: sci.engr.control

Article #241

#241

From: Alexander Bähr
Date: Thu, 02 Aug 2001 08:38

29 lines
2345 bytes

Hello everybody,

I have a question on the static and adaptive kalman filter.

"static kalman filter", as I understand it, is a Luenberger Observer
design using the Riccati formula. You give the covariance matrices for
system noise and measurement noise, and get the optimal feedback gains.
In Matlab: lqe, dlqe

The adaptive kalman filter (which is usually called the kalman filter)
an adaptive algorithm that changes the observer feedback gain at each
step according to some strange formulas- but it still needs given
covariance matrices.

(1) In some experiments with the standard Kalman filter in Matlab, I got
a constant feedback gain after a short while, which depended not on the
input data, but only on the given covariance matrices. It even remained
the same when I "feeded" the filter with nonsense data. Can this be
true?

(2) What is the advance that the adaptive Kalman filter brings? I
thought it was able to find the covariances itself - is there such an
algorithm? Finding covarance matrices by imagination is a difficult
thing :-).

Please answer by mail, I will post the answers.
Thank you in advance.
Alexander

Technical Details

Message-ID: <3B691173.27D37200@srt.tu-darmstadt.de>

Path:

arc1-ca.pugleaf.net!archiver1.google.com!newsfeed.google.com!newsfeed.stanford.edu!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!nntp.infostrada.it!fu-berlin.de!news.tu-darmstadt.de!not-for-mail

Show Full Headers

Complete Article Headers

X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 1117cc,b1abb766cfbb532f,start
X-Google-Attributes: gid1117cc,public
X-Google-Thread: fdd91,b1abb766cfbb532f,start
X-Google-Attributes: gidfdd91,public
X-Google-Thread: 107905,b1abb766cfbb532f,start
X-Google-Attributes: gid107905,public
X-Google-Thread: 104540,b1abb766cfbb532f,start
X-Google-Attributes: gid104540,public
X-Google-ArrivalTime: 2001-08-02 01:40:08 PST
Path: arc1-ca.pugleaf.net!archiver1.google.com!newsfeed.google.com!newsfeed.stanford.edu!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!nntp.infostrada.it!fu-berlin.de!news.tu-darmstadt.de!not-for-mail
From: Alexander Bähr <abaehr@srt.tu-darmstadt.de>
Newsgroups: sci.engr.control,fj.engr.control,fido.ger.control,alt.control.theory
Subject: question: kalman filter
Date: Thu, 02 Aug 2001 10:38:11 +0200
Organization: TU Darmstadt - Institut fuer Stromrichtertechnik
Lines: 32
Message-ID: <3B691173.27D37200@srt.tu-darmstadt.de>
NNTP-Posting-Host: pc055.sr.e-technik.tu-darmstadt.de
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.7 [de] (WinNT; I)
X-Accept-Language: de