9/6/2014 12:31 PM | |
Joined: 9/23/2005 Last visit: 8/27/2024 Posts: 4347 Rating:
|
Hi, I'm interesting also to control pumbs (not with fuzzy, but with artificial inteligence algorithms). Unhapply I don't know any lib read to use for it, and I don't have in short time to develop it. However, you can build one lib, in SCL language. Try to find any example in OSCAT (www.oscat.de/). If you found any sollution, please tell us. |
Denilson Pegaia |
|
9/6/2014 9:14 PM | |
Posts: 5225 Rating:
|
Hi there. Wikipedia [LINK] gives a good descripion of the consept, in my opinion. To my understanding it goes about releative control. In your case you can have x5 conditions:
Phases 2 and 4 can be broken down in smaller increments also. Essentially you could approach this with a ruled based concept that has an update rate. For example we are measuring flow and every 10sec we look at the situation:
Now for example we have Q0.0 as an output making more open. This preset time would then open the valve slightly. After several 10sec refresh rates you will be in place where you want to be. Remember that you could have several preset times to expand phase 2 (and 4) to smaller groupings. If you keep all variables adjustable in a DB, you will be able to tune the system until your are happy with it. That is the general idea. w LATE NOTE.. Sorry I see now that I am talking of a valve and you asked about a pump. Use the above for the base consept. The next post will give clue on using pumps. |
Last edited by: William B. at: 9/6/2014 9:17 PMlate note |
|
9/6/2014 9:29 PM | |
Posts: 5225 Rating:
|
..for pumps: Instead of keeping Q0.0 active for a time, the setpoint of the pump speed can be modified. So if a small jump is needed the setpoint can be added with a small (relative) number. In the case of the extreems 1 and 5 phases, the setpoint is set to maximum or minimum allowed value. Again you can have phases 2 and 4 expand to other groupings. An alternative would be to make a mathematical model that can be used: setpoint change := difference x multiplication factor. The idea is that the setpoint change is always less than you really require, so the system will zoom into the required amount you need. The update rate can also be dinamic, but make sure the flow (and pump speed) has stabilised before updating to a new setpoint change. @Pegaia:
artificial inteligence means different things to different people. Even the control system community is depbating the concept.. I am assuming you refer to the text book explaination as process control community understands it. Combining Fuzzy Logic with a mathematical model and building up a type of memory to modify the mathematical model (adapting multiplication factor for this simplified method) might qualify to a fake AGI.. depending on what the underlying deffinition is. Greetings. w |
Last edited by: William B. at: 9/6/2014 9:31 PMnotes expanded |
|
Follow us on