Newsgroups: comp.ai.fuzzy
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!agate!darkstar.UCSC.EDU!news.hal.COM!decwrl!netcomsv!netcom.com!dfuess
From: dfuess@netcom.com (David A. Fuess)
Subject: Re: Fuzzy control-illusion or future
Message-ID: <dfuessD149tu.4rH@netcom.com>
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
References: <3d3928$f8@cstatd.cstat.co.za> <3d632j$1rm@josie.abo.fi>
Date: Tue, 20 Dec 1994 16:01:06 GMT
Lines: 82

In article <3d632j$1rm@josie.abo.fi>, Johan Pensar RT <jpensar@abo.fi> wrote:
>In article f8@cstatd.cstat.co.za, jorglalk@iaccess.za (Jorg Lalk) writes:
>> The advantage of the fuzzy system is 
>> also the ease with which it can be "tuned" to suit a particular application.
>
>Really? As far as I know, fuzzy controllers are easily tuned for simple
>processes where the human logic is successful, but for a more complex
>process, or a MIMO process, the human logic is mostly insufficient.
>In my experience, people may have problems understanding how to control
>a SISO system with inverse responce, not to mention a high-order MIMO 
>system with much interaction and directionality. 

Berenji showed that a fuzzy system can train itself very quickly (5 
iterations) to perform that which most people cannot do easily. So it is 
not absolutely necessary to have a human expert. You only need to be able 
to recognize success.

>
>I would also not call tuning a fuzzy controller with genetic algorithms 
>"easy". What is the advantage of the fuzzy logic here? The fuzzy logic 
>is in this case used only as a nonlinear function approximation, and
>You could as well tune another nonlinear function and obtain equal 
>performance.
>
>My own interpretation is that, no, fuzzy control is no illusion, but an 
>approximative method used even long before Zadeh even wrote his first papers 
>on fuzzy logic (See e.g. the short letter "Get the fuzz out" by Sergio 
>del Pozo in IEEE Spectrum 1992, sorry don't know the exact number). 
>It is, however, a much more limited method than its proponents are willing 
>to admit. 

I strongly disagree. You cannot say that a discipline is limited until all 
avenues have been explored. In fact many advances in science and 
technology are seen as novelties by the short sighted establishment at 
the time of its birth. The real value is seen and developed by a later 
less biased generation (less short sighted).

>
>It is also something that has got more than it's share of public attention,
>but as expert systems and neural networks who both have had a similar 
>development cycle, fuzzy control will fade to a more modest size. 
>In ten years from now, it may be a technique teached as an alternative 
>to nonlinear PID control, but it will not be something that revolutionized 
>the process control theory. 
>Then as now, Linear Optimal Control by Anderson and Moore from 1971 will be 
>a standard book, probably joined by some of the research done now on 
>nonlinear control.

The above is proof of my assertions. It says "I don't care what you say, 
what you do, what you have, I'm sticking to the old school. 50 years from 
now the archaic methods of control taught and used today will be looked 
upon as we view the technology of 500 years ago (the factor of 10 
accounts for greater than linear, perhaps exponential, increase in 
development) quaint, but dated. Will fuzzy control play a part in that? 
Without a doubt. Because it recognizes the one thing that the 
determinists cannot grasp, that nature cannot and will not ever exactly 
match our precise mathematical models of it, that the differences are 
fundamental and not a function of the exactness of the model, and that 
the differences are, in the end, the cause of the failure of the precise 
model to predict natural behavior.

>
>Johan
>---
>-------------------------------------------------------------------------
>Johan Pensar                                 E-mail: jpensar@abo.fi
>Process Control Laboratory                   Phone:  +358 21 654743
>Abo Akademi University                       Home:   +358 21 382992
>Biskopsgatan 8                               Fax:    +358 21 654479  
>FIN-20500 Abo                                http://www.abo.fi/~jpensar/
>Finland                                      Office: 366 (Axelia)
>-------------------------------------------------------------------------
>           "Unix is inconsistant and power user friendly."
>
>


-- 
 _|_                                                     _|_ 
(___)- - - - - - - - - - - - - - - - - - - - - - - - - -(___)
 |/|        dfuess@netcom.com | dave@myst.llnl.gov       |/| 
(___)- - - - - - - - - - - - - - - - - - - - - - - - - -(___)
