National Instruments 370757C-01 用户手册

© National Instruments Corporation

The system matrix can be calculated using the 

afeedback( )

 function for 

different values of K. Consider two cases: K = 1 and K = 5. 

P = 1/makepoly([1,0],"s")

P (a transfer function) =

  1

  --

  s 

  System is continuous

K1= 1/makepoly(1,"s")

K5= 5/makepoly(1,"s");

Sys1 = afeedback(P,K1);

Sys5 = afeedback(P,K5);

The effect of the value of K on closed-loop performance can be investigated 
using 

perfplots( )

. 

sv1 = perfplots(Sys1,1,1);

Overlap plots:

sv5 = perfplots(Sys5,1,1,{!graph});

for i = 1:4

 plot(sv5(1,i),

{graphnumber=i,line_style=2,keep})?

endfor

In Figure 3-5, you can see that over the bandwidth of 0.1 Hz, the controller 
K = 5 has better regulation (e/d is smaller for K = 5 than for K = 1, with e/n 
about the same for both cases) but uses slightly more actuator effort. Above 
the bandwidth of 0.1 Hz, the e/n and u/n show that the K = 5 controller is 
more sensitive to sensor noise. In classic terms, the K = 5 controller has a 
higher bandwidth.