Hi Everyone,
while designing a controller based on measured frequency response data I
noticed some strange results by in control.nyquist_plot().
Looking at the code I noticed that it first calculates the magnitude and
the phase of the system response and then converts it back to the x and
y components.
I have modified control.nyquist_plot() to directly use the complex
frequency response for plotting:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Index: src/freqplot.py
===================================================================
--- src/freqplot.py (revision 293)
+++ src/freqplot.py (working copy)
@@ -227,14 +227,12 @@
#TODO: Add MIMO nyquist plots.
raise NotImplementedError("Nyquist is currently only
implemented for SISO systems.")
else:
- # Get the magnitude and phase of the system
- mag_tmp, phase_tmp, omega = sys.freqresp(omega)
- mag = np.squeeze(mag_tmp)
- phase = np.squeeze(phase_tmp)
+ # Get the system response as complex numbers
+ z = sys.evalfr(omega)[0,0,:]
# Compute the primary curve
- x = sp.multiply(mag, sp.cos(phase));
- y = sp.multiply(mag, sp.sin(phase));
+ x = np.real(z)
+ y = np.imag(z)
if (Plot):
# Plot the primary curve and mirror image
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
With this patch my nyquist plots look ok. What do you think of this change?
Best regards
Erik Fleischer
--
Andrew Lewis:
"Wenn Sie für einen Dienst nichts bezahlen, sind Sie offenbar nicht
Kundin oder Kunde, sondern die Ware, die verkauft wird."
|
