Data processing: artificial intelligence – Fuzzy logic hardware – Digital fuzzy computer
Reexamination Certificate
1997-08-01
2001-07-03
Davis, George B. (Department: 2122)
Data processing: artificial intelligence
Fuzzy logic hardware
Digital fuzzy computer
C706S001000, C706S009000, C706S900000
Reexamination Certificate
active
06256617
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to a fuzzy cell structure for an electronic processor operating in a fuzzy logics mode.
More particularly, the invention relates to a processor of the type which comprises a fuzzyfication unit receiving a plurality of input variables on its input and being adapted to compute a membership value of such variables in a membership function, a processing unit connected downstream of the fuzzyfication unit to produce a fuzzy set, i.e. the results of a fuzzy logics inference operations performed on said plurality of variables, and a defuzzyfication unit operative to translate said inference results into a so-called crisp value.
2. Discussion of the Related Art
As is well known, electronic processors are currently available which operate in a general purpose fuzzy logics mode, and are sold under the trade name WARP 1-2-3.
These processors are directed to solve model and control problems in processes of significant complexity characterized by having several input variables, a few control variables, a particularly high dynamic range, and precision specifications which allow some latitude.
However, there are situations where these processors are not utilized to full capacity, such as where the number of the variables involved is small. In other situations, by contrast, these processors cannot be used due to their inherently inadequate speed and precision; this being the case, for instance, with the processing of digital signals and with the control problems brought about by a high dynamic range or the need for more accurate control actions.
For a better understanding of the invention aspects, the main rules of operation of the current-generation of fuzzy processors will be reviewed briefly. These processors are capable of processing rules of the following kind:
IF X
1
is A
ij
and X
2
is A
2h
and . . . THEN Y is C
or variations of this same rule wherein the decision variables X, i.e. those appearing in the IF part or portion of the rule, are backed by a sub-set U
i
⊂R, A
ij
being one-dimension terms defined in U
1
.
The structure of these fuzzy processors can be represented schematically by three cascaded blocks, as shown in the appended FIG.
1
.
A first “Fuzzyfication Unit” block is to translate, into linguistic terms, so-called ‘crisp’ observations of the input variables X
i
by computing their membership value at each term. The membership values thus obtained are used by the “Computational Unit” to produce a fuzzy set, or term, representing the linguistic inference of the set of computed rules.
Rule processing is over once the linguistic inference is translated into a crisp value by a “Defuzzyfication Unit” block.
Further features of prior art fuzzy processors can be learned from European Patent Applications No. EP96830173.9 and EP96830090.5 by the Applicant, which are incorporated herein by reference.
An object of the invention is to provide a compact-size dedicated processor which has such structural and functional features that it can solve control problems involving few variables and still be sufficiently fast and accurate for use in the processing of digital signals.
This would allow certain limitations of conventional processors to be overcome.
SUMMARY OF THE INVENTION
The solution idea behind the invention is of making an appropriate choice of the fuzzy model to be used, that is of the type of rules to be used, and above all, of the type of membership functions.
With the intention of placing no limitations on the number of rules to be used in a given system, since computing complexity would be dependent in any case on the number of rules, we have considered acting on the number of rules that are active each time, and identifiable by their having a non-zero degree of activation. In this way, the number of rules can be kept always relatively small, and the activated rules are always identifiable within the rule data base.
On the basis of the above solution idea, we have been able to find a model having the following properties:
1) the sum of the membership functions through the universe of discourse is unity;
2) at any point of the universe of discourse, there are only two membership functions with a non-zero degree of membership;
3) the only aggregation operator used is the AND operator.
Differently from what a quick look into the matter may suggest, the conditions for the use of this model are not constrictive. The resulting fuzzy model is indeed complete, and offers several advantages to be specified hereinafter.
In accord with the above solution idea, the technical problem is solved by a processor as previously indicated and comprising a first memory device connected to the fuzzyfication unit and containing the set of membership functions, a second memory device connected to the defuzzyfication unit and containing the set of crisp value which appear in the THEN part of the fuzzy rules, and a further memory device connected to the defuzzyfication unit.
The features and advantages of the fuzzy processor of this invention will be apparent from the following description of an embodiment thereof, given by way of illustration and not of limitation with reference to the accompanying drawings.
REFERENCES:
patent: 5165011 (1992-11-01), Hisano et al.
patent: 5249258 (1993-09-01), Hisano et al.
patent: 5268847 (1993-12-01), Potetz et al.
patent: 5280624 (1994-01-01), Ikeda
patent: 5371832 (1994-12-01), Eichfeld et al.
patent: 5574826 (1996-11-01), Russo et al.
patent: 5600757 (1997-02-01), Yammamoto et al.
patent: 5621860 (1997-04-01), Russo et al.
patent: 5633986 (1997-05-01), Rizzotto et al.
patent: 5657427 (1997-08-01), Rizzotto et al.
patent: 5724483 (1998-03-01), Gianguido et al.
patent: 5737493 (1998-04-01), Viot et al.
patent: 5751908 (1998-05-01), Madau et al.
patent: 5790755 (1998-08-01), Pagni et al.
patent: 5799132 (1998-08-01), Rizzotto et al.
patent: 0544629 A2 (1993-06-01), None
patent: 0798631 A1 (1997-10-01), None
patent: 0831394 A1 (1998-03-01), None
Surmann, H.; Ungering, A.; “Fuzzy Rule-Based Systems on General-Purpose Processors”; IEEE Micro; vol. 15, No. 4, pp. 40-48, Aug. 1995.*
Basch, D.; Zagar, M.; “Processor Architecture Model for Fuzzy Control”; Proceedings of the 1995 IEEE International Symposium on Intelligent Control; pp. 189-194, Aug. 1995.*
“Fuzzy Logic Single-Chip Solution”; Oki Electric Industry Co., Ltd.; 8 pages, Apr. 1993.*
Chieuh, T.; “Optimization of Fuzzy Logic Inference Architecture”; Computer; vol. 25, No. 5, pp. 67-71, May, 1992.*
Watanabe, H.; Symon, J.; Dettloff, W.; Yount, K.; “VLSI Fuzzy Chip and Inference Accelerator Board Systems”; Proceedings of the 21st International Symposium on Multiple-Valued Logic; pp. 120-127, May, 1991.
Cuce' Antonino
Pelos Enrico
Davis George B.
Galanthay Theodore E.
Morris James H.
SGS--Thomson Microelectronics S.r.l.
Wolf, Greenfield & Sacks, P.C
LandOfFree
Fuzzy logic electronic processor does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Fuzzy logic electronic processor, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Fuzzy logic electronic processor will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2442201