Group Abstract Group Abstract

Message Boards Message Boards

Question on discrete time model


In designing a control system in the continuous frequency domain, and then converting it to a discrete time system, I get to the result below.
 In[24]:= discreteClosedLoop =
  ToDiscreteTimeModel[closedLoop, .5, z] // FullSimplify
 Out[24]= TransferFunctionModel[{{{(
      3.1886932013146107`*^25 + z (-4.875667541784289*^25 + z (-2.\
 668411863050167*^25 + 5.395948880048733*^25 z))) \[Alpha]}}, {{
    9.419520692765025*^25 (-1. + 1. z)^3 + (
       3.1886932013146004`*^25 + z (-4.875667541784272*^25 + z (-2.\
 6684118630501584`*^25 + 5.395948880048714*^25 z))) \[Alpha]}}}, z,
SamplingPeriod -> 0.5, SystemsModelLabels -> {{None}, {None}}]

It's a bit hard to read in that view. Here is a screen capture:

First, it appears there is a factor of at least 10^25 that can be extracted from both the numerator and denominator and cancelled. Does anyone know how I wind up with that, and what to do about it?

Second, is there a way to get the polynomial in z out of the model, so it can be worked with using InverseZTransform? "Normal" does not extract it, like it would for other encapsulations in Mathematica.

Thanks and best regards,

POSTED BY: David Keith
10 months ago
 In[34]:= discreteClosedLoop = TransferFunctionModel[{{{(3.1886932013146107`*^25+
       z (-4.875667541784289*^25+z (-2. 668411863050167*^25+5.395948880048733*^25 z))) \[Alpha]}},
     {{9.419520692765025*^25 (-1.+1. z)^3+(3.1886932013146004`*^25+z (-4.875667541784272*^25+
       z (-2. 6684118630501584`*^25+5.395948880048714*^25 z))) \[Alpha]}}},
 (* Pull out the numerator and denominator using the Part function. *)
 In[35]:= num = discreteClosedLoop[[1,1,1,1]]
 In[36]:= denom = discreteClosedLoop[[1,2,1,1]]

In[37]:= Expand[num * 10.^-25] / Expand[denom * 10.^-25]
Out[37]= (3.18869 \[Alpha]-4.87567 z \[Alpha]-1.33682*10^15 z^2 \[Alpha]+5.39595 z^3 \[Alpha])/(-9.41952+28.2586 z-28.2586 z^2+9.41952 z^3+3.18869 \[Alpha]-4.87567 z \[Alpha]-1.33682*10^16 z^2 \[Alpha]+5.39595 z^3 \[Alpha])

Not super elegant, but does what you want.

Functions such as FactorTerms choke on the floating-point (i.e., approximate) numbers.

You can give this to InverseZTransform,
InverseZTransform[%, z, n]
The output is large.

As for where the 10^25 factors come from, a Control Systems whiz would need to see your data.
POSTED BY: Bruce Miller
10 months ago
Another way is as follows
  • The given transfer function 
discreteClosedLoop = TransferFunctionModel[{{{(3.1886932013146107`*^25+
       z (-4.875667541784289*^25+z (-2. 668411863050167*^25+5.395948880048733*^25 z))) α}},
     {{9.419520692765025*^25 (-1.+1. z)^3+(3.1886932013146004`*^25+z (-4.875667541784272*^25+
       z (-2. 6684118630501584`*^25+5.395948880048714*^25 z))) α}}},
  • Extract the discrete transfer function:
  • Perform the simplification:
you could also remove the braces, as Bruce did, using the part function([[1,1]]).
POSTED BY: Ahmed Mogahed
10 months ago
Thank you both. Just what I needed!
POSTED BY: David Keith
10 months ago