|
|
How to Use BESTune?
(More Info: Go
back to BESTune home page)
BESTune is very easy to use. Just run the executable "BESTune.exe"
and follow the self-explanatory instructions on the GUI or the step-by-step
instructions below.
Step 1: Create data file co_pv.txt
Create a file named co_pv.txt
which contains the open-loop bump test data (or closed-loop data if open-loop
data is unavailable). It is important that the CO and PV be correctly chosen as
defined by the PID equations and the CO and PV data show
enough dynamics of the process. This drawing shows
where to find the CO and PV variables in a typical PID control loop. An
example of where to find CO and PV is given as a demo in http://bestune.50megs.com/piddesign.htm.
The structure of co_pv.txt is as follows:
co1 pv1
co2 pv2
co3 pv3
: :
Click this
link to open such a co_pv.txt file.
Explanation:
|
Time |
CO samples |
PV samples |
Explanation |
|
t1 |
co1 |
pv1 |
At time t1, co1 and pv1 are sampled |
|
t2 |
co2 |
pv2 |
At time t2=t1+T, co2 and pv2 are sampled |
|
t3 |
co3 |
pv3 |
At time t3=t2+T, co3 and pv3 are sampled |
|
: |
: |
: |
So on so forwards |
The sampling periods are the
same T. T should be close to, the sampling period used by your PID equation
(i.e. the time period that your PID controller updates its output). However,
if the ratio
(process dead-time)/(sampling period)
is larger than 5~9, you have three
options to make this ratio less than 2~5 (1~2 is recommended).
In some applications you are
not allowed to have options 1 and 2. But option 3 is usually enough to reduce
the ratio.
It is important that your CO
and PV data should show how changes in CO will affect the PV variable. For this
purpose it is suggested that the number of samples be no less than 200~300
(1000~3000 samples is recommended).
Don't leave any extra blanks at
the end of the co_pv.txt file.
Step 2: Enter sampling period T used in
"co_pv.txt" in seconds.
Step 3: Enter your process' dead-time in
seconds.
The dead-time, also called time
delay or lag, is the time period from the time when there is a change (increase
or decrease) in CO to the time when PV begins to respond to this CO change.
Step 4: Make a guess of the order of your
process. If you have no idea about this, please select 2nd order. This is a
good choice for most processes.
Step 5: Select "PI" or "PID" if you want your
controller to be a PI or PID controller
If your PID controller is of
type A or type B, it is recommended that you select "PI" to make your
controller a PI controller.
Step 6: Set tightness for your controller.
Move the slide to the right if
you want your controller to be tighter. The tightness index can change
from 1 to 100. If your PID controller is of type A or type B, it is recommended
that you reduce the tightness index to under 20. For more details take a look
at http://bestune.50megs.com/typeABC.htm.
Step 7: Click the step 7 push button to tune
your PI or PID controller.
The time needed for tuning
increases as the ratio of the dead-time (step 3) to the sampling period
(step 2) increases. Typically the tuning is finished in less than 1 second.
Please wait until the top bar of the BESTune GUI turns blue.
Step 8: Select the PID controller you are
using from the pull down menu.
If your controller is not in
the menu, you can choose one that is closest to your controller and
convert the tuning result to that of yours. If you want your controller to
be included in the list, please tell us what PID equation your controller
is using.
Step 9: Click the step 9 push button to show
the tuning results for your controller.
Please pay attention to the
units of the tuning constants. The progress bar at the lower and left corner of
the GUI shows how much dynamics is in your co_pv.txt file. High dynamics in
your co_pv.txt file is important for getting correct tuning results and hence
is an important criterion when choose your CO and PV data.
Step 10: Select the PID equation type your
controller is using.
There are three PID equation
types:
Type A: All the three P, I, and D terms act on
the error e=setpoint-PV. This is the so called textbook PID equation. We don't
recommend the use of type A.
Type B: The P term and I term act on the error
e=setpoint-PV and the D term act on PV. Most industrial PID controllers (e.g.,
Allen-Bradley's PLCs) use this type of PID equation. Type B is better than Type
A.
Type C: Both P and D terms act on PV, only the
I term acts on the error e=setpoint-PV. We strongly recommend this type of PID
equations. If you are allowed to implement your own PID equation, it is
strongly recommended that you implement type C PID equation. Click here to find more about
these three types.
Although we strongly recommend
type C PID equations, if your controller is not type C, you have to correctly
choose the actual type of your PID equations. It is very sad that most PID
equations implemented in industry are either type A or type B. In such a case,
you have to choose type A or type B.
Step 11: Enter setpoint sequence, upper and
lower limits for CO, and simulation time span.
Step 12: By clicking on the push button, you
can do the off-line "what if" simulation.
This allows you to see how your
PID controller will perform if the tuning constants given by BESTune are
used. If your PID controller is of type C, your controller's performance
is usually very good. However, if your PID controller is of type B and
especially of type A, you may observe overshoot in the "what if"
simulation. If this occurs, you can reduce the overshoot by reducing
the controller tightness in step 6 (go back to step 6 and reduce the
tightness until no overshoot occurs at Step 12). Click here to find more about
these three types.
Step 13: If you wish, you can try tuning your
controller manually.
Step 14: Click this push button to see the
performance of your manually tuned or existing control loop.
Step 15: Click here to get the performance
index of your manually tuned PID loop.
More info: Go to BESTune home page
