Fuzzy PID

[andrews]

Has anyone implemented a Fuzzy logic PID controller on a PIC???

[valemike]

I only did very very few PID controllers, but here is a good starting point.

Google the following terms:
PID PHD embedded.com

It will lead you to a year 2000 article Embedded Systems Programming magazine entitled "PID Without a PHD". Reading it will give you a usable ~10 lines of code which makes the heart of your PID control source code, that you can use on a PIC. (Replace the "doubles" with "floats")

As far as fuzzy logic goes, i thought "fuzzy" and "PID" are apples and oranges. What are you trying to control anyways? Temperature? Motor Speed?

[andrews]

Thanks,

I'll look up that link this afternoon.

Will be using it in a couple of applications - controlling current in an RTD in a hot wire anemometer application to acheive a mass flow (air) and also to control a I/P converter to control pnematic valves. Essentailly in the end a mass flow controller

My initial understanding is that a fuzzy PID controller is much smoother than a traditional PID controller. I first stumbled across an instance of it here http://www.bytecraft.com/downloads/fuzlogic.pdf and have since been poking around google this morning trying to find more info.

Like everyone else, I want a controller that is fast and stable - trying to answer those questions now and use what is somewhat state of the art but not bleeding edge..

[SherpaDoug]

Didn't fuzzy logic die of its own weight back before 2000? It seemed to be a complicated and resource intensive solution for problems that didn't really exist. It excelled in situations where there were important unknowable variables, such as heating a tank of water at an exact temperature curve when you could not even guess the amount of water or the heater power. Such applications are mathematically interesting, but in real life nonexistant.

It was a useful buzzword for Marketing until they wore it out.
_________________
The search for better is endless. Instead simply find very good and get the job done.

[jma_1]

Either method of closed-loop control works. How well each works and development time are another matter. There is a tool kit for PICs from fuzzytech ( http://www.fuzzytech.com/ ) that I've heard good things about. Another fuzzy option is a tool called FUDGE which Motorola came out with years ago. The tool can output ANSCI C code (which would more than likely require porting over to PIC). One good and bad feature of Fuzzy logic is flexibility. If your control has a specific behavior of XXX which is characterized by YYY, it's easy to come up logic to control or account for the behavior. Defining tangible measurements and limits for the fuzzy control may take time, and are not easily duplicated on a benchtop.

Justin