Re: Fuzzy Control

From: P. Sarma (pramits@vsnl.com)
Date: Sun Dec 31 2000 - 13:54:28 MET

  • Next message: fuzzy-owner@dbai.tuwien.ac.at: "NEWER FUZZY-MAIL MESSAGES ..."

    One of the keys to good fuzzy logic control (FLC) is knowing that it can in
    fact replace PID's. One of the keys
    to it replacing PID's is the less widely known thing that all FLC's are in
    fact a form of nonlinear PID. It is less widely known in a "public" sense:
    it is very well known from the published literature. In fact, given inputs
    as error e, change-in-error (or de/dt) it is easy to show that, along with
    some standard form of intermediate FLC processing (fuzzy operators,
    rule-bases (FRB), defuzzifying operators), it directly leads to what are
    often called fuzzy PID's (FPID's). This should not be confused with the
    fuzzy-tuned PID's, where the {P,I,D} parameters are tuned online by some
    fuzzy block.

    The FPID so strongly resembles a PID because that is the way it was created
    and designed by Mamdani in 1974. Clearly, the best known, and highly
    successful, minimalist input/output controller known earlier was the crisp
    PID, making it an ideal candidate for evolving a new format. Therefore, a
    linguistic or fuzzy representation of this crisp linear structure would lead
    directly to the fuzzy PID. Given this, it is not so difficult to construct
    an appropriate FLC processor, particularly the central FRB. This linguistic
    or fuzzy version permits the construction of a vast number of variations on
    the same theme of a FPID = nonlinear PID, by the original Mamdani design.

    The performance depends on the designer's choice of a large number of
    "tuning parameters": fuzzy partitions, fuzzy set geometry, FRB,
    (de)fuzzifying operators, and fuzzy logical operators. All of these go to
    form a tailorable nonlinear controller, or FPID. It has already been proven
    that the Mamdani FLC (M-FLC) forms a Universal Function Approximator (UFA),
    which is a property shared by the related TSK-FLC. A direct practical
    consequence of this result is that any desired closed-loop FLC response is
    achievable, given an appropriate FLC design. The theoretical possibility of
    this itself transcends the root crisp linear PID's abilities, and provides
    the fundament on which modern FPID's are designed.

    Pramit

    -----Original Message-----
    From: james.becker.b@bayer.com <james.becker.b@bayer.com>
    To: Multiple recipients of list <fuzzy-mail@dbai.tuwien.ac.at>
    Date: Sunday, December 31, 2000 9:23 AM
    Subject: re: Fuzzy Control

    >There are a few industrial practitioners who search this newsgroup.
    >
    >There are a few people who have done fuzzy logic controllers which
    >mimic the response of PID controllers. But one has to question the
    >reason for doing this. Why put in a great new V-8 engine and then
    >adjust it to respond like a VW Beetle? An old Beetle.
    >
    >The best way to implement a fuzzy logic controller is to look at the
    >system response to discover which other inputs can effect the
    >controller you are creating. Include these inputs in logic and you
    >can create a multivariable controller. A typical PID replacement
    >fuzzy logic controller uses the deviation from set point (SP-PV) and
    >the derivative of the PV as inputs.
    >
    >There is so much more an engineer can do to adjust the response
    >of a fuzzy logic controller to the specific system. It is possible to make
    >the controller respond quickly and have little or no overshoot. This is
    >accomplished by adjusting the controller response in individual
    >matrix blocks or the membership functions.
    >
    >Also, the engineer can create the equivalent of feedforward controls
    >or any other type of advanced control done with PID controllers.
    >
    >I suggest that you look at the Inform Software website for their product
    >called FuzzyTECH, www.fuzzytech.com. You can download the
    >software for a trial. I suggest that you look at the boiler simulation that
    >comes with the product. You can modify the logic and make the
    >controller respond in almost any way you want it to operate. By the way,
    >the entire User's Manual is included in .pdf format with the download.
    >
    >The company is in Germany, but they have an office in Chicago.
    >
    >Jim Becker
    >
    >
    >
    >Subject: Fuzzy Control
    >From: Gregory L. Hansen (glhansen@steel.ucs.indiana.edu)
    >Date: Wed Oct 11 2000 - 01:19:57 MET DST
    >
    > sorted by: [ date ] [ thread ] [ subject ] [ author ]
    > Next message: robert_wilhelm_land: "Re: Grid-partitioning Fuzzy
    Systems"
    > Previous message: Jerry Avins: "OT Re: Visual computation using
    aliasing
    >and interference"
    >
    >
    >
    >Is this the right place to ask about fuzzy logic in control systems? It's
    >the only fuzzy logic newsgroup I could find.
    >
    >I'm trying to understand fuzzy logic control systems compared to more
    >traditional systems like the PID controllers. The examples go through
    >decision matrices and elaborate logic and calculation to do what seems to
    >be the moral equivalent of "if sensor moves away from setpoint, adjust
    >parameter in a direction that will bring the system back", which seems
    >equivalent to the integral term in a PID controller. What does fuzzy gain
    >you?
    >
    >It's been said that fuzzy logic lets you use simple English rules and
    >focus on what you want the system to do rather than trying to model the
    >system, and that it easily handles non-linear systems, in contrast to
    >control theory. And I suppose that's true to a point. But even a simple
    >PID can handle systems that are non-linear to a degree, because all it
    >does is produce an output related to the deviation of the system from the
    >ideal, which is pretty much focusing on what you want the system to do,
    >and there are simple and even automated methods to calibrate the run
    >parameters. And in a fuzzy system you still need to define control
    >regions -- number, locations, widths, overlaps. You still need to
    >defuzzify the output, you need to tweak parameters. It seems like those
    >simple English statements are just code words for the modeling and
    >mathematics that must eventually be done. And I'd be surprised if fuzzy
    >controllers never suffer instabilities.
    >
    >I like fuzzy logic. It looks fun and interesting, and probably very
    >useful in things like expert systems. But I feel a little cheated in how
    >I've seen fuzzy control systems explained. The popular media seems to
    >pretend that all non-fuzzy systems are stupid on/off affairs and that
    >control theory doesn't exist. They bemoan the fact that the U.S. hasn't
    >been putting many resources into fuzzy logic, but don't mention if that
    >might be because U.S. industry is satisfied with traditional controllers.
    >The more technical articles compare traditional controllers in fuzzy logic
    >terms, like the numerous if...then statements a PID controller must have
    >rather than just giving a discretized formula that it follows relentlessly
    >with no decision element whatsoever; a formula, I might add, that seems
    >simpler than determining the membership of an input to various sets,
    >cutting out sections of an output set, and finding the centroid. I feel
    >like a lot of hype surrounds the field.
    >
    >--
    >"A good plan executed right now is far better than a perfect plan
    >executed next week."
    > -Gen. George S. Patton
    >###########################################################################
    #
    >This message was posted through the fuzzy mailing list.
    >(1) To subscribe to this mailing list, send a message body of
    >"SUB FUZZY-MAIL myFirstName mySurname" to listproc@dbai.tuwien.ac.at
    >(2) To unsubscribe from this mailing list, send a message body of
    >"UNSUB FUZZY-MAIL" or "UNSUB FUZZY-MAIL yoursubscription@email.address.com"
    >to listproc@dbai.tuwien.ac.at
    >(3) To reach the human who maintains the list, send mail to
    >fuzzy-owner@dbai.tuwien.ac.at
    >(4) WWW access and other information on Fuzzy Sets and Logic see
    >http://www.dbai.tuwien.ac.at/ftp/mlowner/fuzzy-mail.info
    >(5) WWW archive:
    http://www.dbai.tuwien.ac.at/marchives/fuzzy-mail/index.html
    >
    >
    >
    > Next message: robert_wilhelm_land: "Re: Grid-partitioning Fuzzy
    Systems"
    > Previous message: Jerry Avins: "OT Re: Visual computation using
    aliasing
    >and interference"
    >
    >
    >
    >This archive was generated by hypermail 2b25 : Wed Oct 11 2000 - 01:22:48
    MET
    >DST
    >
    >
    >###########################################################################
    #
    >This message was posted through the fuzzy mailing list.
    >(1) To subscribe to this mailing list, send a message body of
    >"SUB FUZZY-MAIL myFirstName mySurname" to listproc@dbai.tuwien.ac.at
    >(2) To unsubscribe from this mailing list, send a message body of
    >"UNSUB FUZZY-MAIL" or "UNSUB FUZZY-MAIL yoursubscription@email.address.com"
    >to listproc@dbai.tuwien.ac.at
    >(3) To reach the human who maintains the list, send mail to
    >fuzzy-owner@dbai.tuwien.ac.at
    >(4) WWW access and other information on Fuzzy Sets and Logic see
    >http://www.dbai.tuwien.ac.at/ftp/mlowner/fuzzy-mail.info
    >(5) WWW archive:
    http://www.dbai.tuwien.ac.at/marchives/fuzzy-mail/index.html
    >
    >

    ############################################################################
    This message was posted through the fuzzy mailing list.
    (1) To subscribe to this mailing list, send a message body of
    "SUB FUZZY-MAIL myFirstName mySurname" to listproc@dbai.tuwien.ac.at
    (2) To unsubscribe from this mailing list, send a message body of
    "UNSUB FUZZY-MAIL" or "UNSUB FUZZY-MAIL yoursubscription@email.address.com"
    to listproc@dbai.tuwien.ac.at
    (3) To reach the human who maintains the list, send mail to
    fuzzy-owner@dbai.tuwien.ac.at
    (4) WWW access and other information on Fuzzy Sets and Logic see
    http://www.dbai.tuwien.ac.at/ftp/mlowner/fuzzy-mail.info
    (5) WWW archive: http://www.dbai.tuwien.ac.at/marchives/fuzzy-mail/index.html



    This archive was generated by hypermail 2b30 : Sun Dec 31 2000 - 13:59:08 MET