Searching \ for 'Re[4]: FuzzyTech MP' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=fuzzytech+RE%5D
Search entire site for: 'FuzzyTech MP'.

Truncated match.
PICList Thread
'Re[4]: FuzzyTech MP'
1996\04\08@095253 by Mike Riendeau

flavicon
face
>Mike: fuzzy logic doesn't "know" anything; it would be more useful to
>the debate to limit that sort of personification.

I think I implied that in the second half of the paragraph.

>For that matter, the closed form math doesn't "know" where the poles
>& zeros are, either, no matter what algorithm you choose to
>express it.

Your right, the engineer who analyzed the plant response knows
where the poles and zeros are, and his closed form algorithm
provides the gain and phase compensation.

>And, most importantly, physical processes don't have poles & zeros,
>anyway. The poles & zeros are a feature of your mathematical model of
>the process.

I don't know where you studied control theory, but If you are using
a linear time-invariant system to control a physical process, It
had better have a response in the frequency domain.

>There is no reason that I know of to believe that there
>are no other valid models of physical processes, or to
>believe that other models might not be as good.

I didn't say there were no other valid models.  I said if you have
a model which in closed form, you know what compensation will
provide stability for all time.

                               Regards,
                                  Mike

1996\04\08@122728 by John Payson

flavicon
face
> Your right, the engineer who analyzed the plant response knows
> where the poles and zeros are, and his closed form algorithm
> provides the gain and phase compensation.
>
> >And, most importantly, physical processes don't have poles & zeros,
> >anyway. The poles & zeros are a feature of your mathematical model of
> >the process.
>
> I don't know where you studied control theory, but If you are using
> a linear time-invariant system to control a physical process, It
> had better have a response in the frequency domain.
>
> >There is no reason that I know of to believe that there
> >are no other valid models of physical processes, or to
> >believe that other models might not be as good.
>
> I didn't say there were no other valid models.  I said if you have
> a model which in closed form, you know what compensation will
> provide stability for all time.

I think one of the issues with fuzzy systems is that real-life systems
seldom exhibit ideal linear behavior; PID systems can be unstable or
conditionally-stable on such systems and changing system behavior can
change controller response.  While fuzzy systems will seldom have closed-
loop behavior as good as a perfectly-tuned PID system, they can often be
made less sensitive to system characteristics.

As a (very) rough analogy, a camera lens which is designed to focus perfectly
at a certain distance will be very sensitive to changes in that distance;
objects at other distances will be out of focus.  By contrast, a lens which
is focuses less well at is focusing distance may blur less at other distances.
The extreme example of this is a pinhole camera which blurs objects at all
distances roughly uniformly.

1996\04\09@101831 by terogers

flavicon
face
John Payson wrote:
> I think one of the issues with fuzzy systems is that real-life systems
> seldom exhibit ideal linear behavior; PID systems can be unstable or
> conditionally-stable on such systems and changing system behavior can
> change controller response.  While fuzzy systems will seldom have closed-
> loop behavior as good as a perfectly-tuned PID system, they can often be
> made less sensitive to system characteristics.

John: you make a good point. But is it true that a fuzzy control system
will seldom have as good a loop behavior as a tight pid system? I
haven't ever seen any theoretical treatment of this.

- Tom

More... (looser matching)
- Last day of these posts
- In 1996 , 1997 only
- Today
- New search...