You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This version requires the Modelica 3.2.1 Library. It is backward compatible to the previous version 2.3.2
Improvements performed in version 2.3.3 Build 1 (Sept. 8, 2015):
New function StateSpace.Transformation.toBalancedForm: Performs a similarity transformation to a balanced form of the linear system. The goal is to find state so that further numerical computations on the system are more reliable.
Improved functions:
ModelAnalysis.Poles
ModelAnalysis.Zeros
StateSpace.Plot.polesAndZeros
System is now balanced, before eigenvalues and transmission zeros are computed
(to improve numerics for systems with largely varying zeros and poles).
By default, the poles/zeros are also printed to the output window (and not only plotted).
Improved functions:
StateSpace.Plot.bodeSISO
StateSpace.Plot.bodeMIMO
System is now balanced, before bode plot is computed
(to improve numerics for systems with largely varying zeros and poles).
The computation of the bode plot is now 1-2 orders of magnitudes faster
for large systems (the implementation of the algorithm was restructured
and part of the restructured functionality is now pre-compiled in object
code using the __Dymola_translate=true annotation.
Optionally, it is now possible to store the frequency, magnitude and angle
vectors of the bode diagram on file.
Improved function:
ModelAnalysis.FullAnalysis
If the system is too large (> 50 states, inputs, or outputs), then the
system is not printed to the output.
The step response is included as png-file in the html log file systemReport.html (if several plots are generated, only the last one is
currently stored on file (needs to be improved in the future).
Bug fixes performed in version 2.3.3 Build 1 (Sept. 8, 2015):
StateSpace.Plot.bodeSISO,StateSpace.Plot.bodeMIMO, ZerosAndPoles.Plot.bode
Defining logX=false, still computed the frequency vector in logarithmic scale
(and f_min=0 gave an error). This has been fixed.
If a system had many zeros and poles with large absolute values,
an overflow could occur. The algorithm has been changed to make
this situation much more unlikely.
Minor bugs fixed (due to stricter checking of Dymola 2016 FD01, warnings had been
present that protected variables in functions have the input/output attributes;
these wrong attributes have been removed.
The following functions are not correct Modelica because local and result arrays depend
on dimensions defined locally (but must be from inputs or parameter expressions).
Since this is not practical to fix, the Dymola specific annotation __Dymola_allowForSize=true was introduced in Dymola 2016 FD01
to allow relaxed rules for Modelica dimension definition: