These tools support design of control systems for your plant. By using simulation and exhaustive search, performance can be computed and a good controller found without simplifying assumptions such as dominant pole pairs.
There are two main files: userSetupTemplate.py and controlDesign447.py.
This where you specify up to four key parts of your problem: The plant, the controller,
the performance specifications, and search parameters.
The best practice is to start by copying the suppled userSetupTemplate.py file
to a new one named userSetupMY_PROB.py (i.e. use the hw problem number in the filename).
userSetupMY_PROB.py
is then specific to your problem.
This file has three "verbs." You will add one verb to the command line:
- Rlocus
- StepResponse
- Optimize this selects how you want to use your design information.
For example, to get a clickable root locus for your controller and plant:
>python3 usersSetupMY_PROB.py Rlocus <ENTER>
is a library of classes and functions that will support your design process. You should not have to edit it at all.
- select the type of controller you want to use (or are required to use). The options are
- 'Kct' - C(s) = K (K>0) ('Constant gain K')
- 'LLC' - C(s) = K (s-pole)/(s-zero) ('Lead-Lag controller')
- 'PID' - C(s) = Kp + Ki/s + Kds or Kd(s+z1)(s+z2)/s ('Proportional-Integral-Derivative Controller')
You enter the 3-letter code as shown in the setup template file.
-
do a manual design to start off with your controller parameters in the right "ballpark". Enter the manual parameters into your setup template file.
-
Identify the region of the s-plane that may have the performance you want.
-
run
>python3 userSetupMY_PROB.py Rlocus
You will see the root locus plot. Click on the point you want and note the gain value (small font on top of the RL plot frame).
-
Close the root locus plot. The software will ask you for that gain value - enter it.
-
Observe the computed step response for your new gain. Is it somewhat close to meeting the specs? if so, go to step 8. if not, go back to step 4 and click a different spot - is the step response better? Settle for the best response you can get with a couple of tries.
-
Once you have something reasonable, you are ready to optimize. Initialize your controller (in the setup file) to your best gain value from step 6.
-
Set the search parameters in your
userSetupMY_PROB.pyfile. (see Lecture and text for detail on this.) Typically you will start with a high range (e.g. 50) and a small number of values (~4) to quickly make sure things are working without errors. Then raisenvalsto say 40-50 and go have lunch. -
With the result of your long search, you can zoom in for fine tuning. Selecting the set of gains you like best, reduce the range to say 1.5 and reduce the
nvalsto get at least 3% change between search steps (smaller is a waste of time). -
Check for hitting range limits, and repeat searches from new starting points to perfect the results.