Advanced system identification techniques for wind turbine structures Page: 4 of 10
This article is part of the collection entitled: Office of Scientific & Technical Information Technical Reports and was provided to Digital Library by the UNT Libraries Government Documents Department.
The following text was automatically extracted from the image on this page using optical character recognition software:
ADVANCED SYSTEM IDENTIFICATION TECHNIQUES
FOR WIND TURBINE STRUCTURES
Jan T. Bialasiewicz
Richard M. Osgood
National Renewable Energy Laboratory
National Wind Technology Center
1617 Cole Boulevard
Golden, CO 80401-3393
The new approach to modal parameter identification,
presented in this paper, uses an asymptotically stable
observer to form a discrete state-space model for a
wind turbine structure. The identification is per-
formed using input-output time-series. A special
software package developed in this research has been
tested using the data generated by the ADAMS'
model of the Micon 65/13 wind turbine structure.
Numerical and graphical presentation of some of the
results, generated by the programs developed, illus-
trates the range of their applicability.
The goal of this research was to develop advanced
system identification techniques to accurately meas-
ure the frequency response functions of a wind tur-
bine structure immersed in wind noise. To allow for
accurate identification, we developed a special test
signal called the Pseudo-Random Binary Sequence
(PRBS). The PRBS signal produces the wide-band
excitation necessary to perform system identification
in the presence of wind noise. The techniques pre-
sented here will enable researchers to obtain modal
parameters from an operating wind turbine. More
importantly, the algorithms we have developed and
tested (so far using input-output data from an ADAMS
model of a wind turbine structure) permit state-space
representation of the system under test, particularly
the modal state-space representation. This is the only
system description that reveals the internal behavior
of the system, such as the interaction between the
physical parameters, and which, in contrast to trans-
fer functions, is valid for non-zero initial conditions.
Sandia National Laboratories' (SNL) Natural Excitation
Technique (NExT) for modal parameter extraction from
operating wind turbines uses the measured system
ADAMS is a registered trademark of Mechenical Dynam-
outputs obtained as a result of natural wind excitation
[21. Generally, the cross-correlation function of such
outputs has the shape of the system's impulse re-
sponse and therefore allows one to extract modal fre-
quencies and damping ratios. SNL's researchers have
done this using one of the system realization algo-
rithms (such as the Eigensystem Realization Algo-
rithm developed at the National Aeronautical and
Space Administration's [NASA] Langley Research
Center). In other words, SNL researchers assume that
the cross-correlation function represents the sequence
of Markov parameters or impulse response of the sys-
tem to be identified. Such an assumption will not lead
to any input-output model of the system, such as a
transfer function or state-space representation. To
identify any of these input-output characteristics, it is
not enough to excite the system with a frequency-rich
signal (such as natural wind noise) but one must also
measure this signal. System identification requires a
frequency-rich input-output history.
Researchers have developed many system identifica-
tion techniques and applied them to state-space
models to identify modal parameters. Most tech-
niques use sampled-impulse system response histo-
ries, known as system Markov parameters. The new
approach, presented here, uses the results obtained
by researchers at NASA's Langley Research Center [3-
5. Rather than identifying the system Markov pa-
rameters (which may exhibit very slow decay), one can
use an asymptotically stable observer to form a stable,
discrete state-space model to identify the system.
The organization of this paper is as follows. In Section
2, we discuss how the excitation or input signal
should be chosen. Then, in Section 3, we present the
problem of making a proper selection of measure-
ments to be obtained from a modal test. In Section 4,
we introduce the Observer/Kalman Filter state-space
model whose identification is performed by the
DISTRdBUTION OF THIS DOUT IS UNiTED
Here’s what’s next.
This article can be searched. Note: Results may vary based on the legibility of text within the document.
Tools / Downloads
Get a copy of this page or view the extracted text.
Citing and Sharing
Basic information for referencing this web page. We also provide extended guidance on usage rights, references, copying or embedding.
Reference the current page of this Article.
Bialasiewicz, J.T. & Osgood, R.M. Advanced system identification techniques for wind turbine structures, article, March 1, 1995; Golden, Colorado. (digital.library.unt.edu/ark:/67531/metadc688470/m1/4/: accessed December 9, 2018), University of North Texas Libraries, Digital Library, digital.library.unt.edu; crediting UNT Libraries Government Documents Department.