Data processing: artificial intelligence – Fuzzy logic hardware – Digital fuzzy computer
Reexamination Certificate
1997-12-01
2001-07-10
Powell, Mark R. (Department: 2122)
Data processing: artificial intelligence
Fuzzy logic hardware
Digital fuzzy computer
C706S008000
Reexamination Certificate
active
06260030
ABSTRACT:
BACKGROUND OF THE INVENTION
Fuzzy controller designs are based on using a knowledge-base (KB) which consists of fuzzy rules and membership functions. To utilize this KB, existing designs of fuzzy controllers employ approximate reasoning mechanisms such as the Compositional Rule of Inference (CRI) which requires the formation of fuzzy matrices. The fuzzy matrices grow in complexity as the number of inputs/outputs of the controller increases, thereby requiring larger memory and more intensive computations. For all input signals of the fuzzy controller to contribute to its output, the scheme evaluates all the rules to make its decision giving rise to high computational requirements and coarser approximation. At the defuzzifier each rule is associated with one membership function (MF) value, which is identified herein as membership grade (MG). The scheme utilizes these MG's to calculate the controller output. To obtain the final output of the controller, all the possible rules are fired giving rise to many outputs, and these outputs are then averaged using special methods.
Due to the above, fuzzy controllers, even though powerful, are difficult to optimize and generally face difficulties in coping with real-time applications for fast and complex multivariable processes.
It can be seen, therefore, that using a scheme which reduces the number of rules that the fuzzy controller requires to fire, as well as employing a defuzzification method whose average is less coarse will result in a fuzzy controller with an improved response, easier to tune and requires much less memory for processing its knowledge-base.
Other relevant literature includes:
1. L. Sultan and T. H. Janabi, “Fuzzy Logic Controller”, U.S. Pat. No. 5,499,319; 1996.
2. L. A. Zadeh, “Outline of a new approach to the analysis of complex systems and decision processes,” IEEE Trans. Systems, Man and Cybernetics, Vol. SMC-3, pp. 28-44, 1973.
3. L. A. Zadeh, “The concept of a linguistic variable and its application to approximate reasoning, I and II”, Inform. Sci., Vol. 8, pp. 199-249 and vol. 9, pp. 301-357, 1975.
4. L. A. Zadeh, “Fuzzy Logic”, IEEE Computer Magazine, pp. 83-93, Apr. 1988.
5. L. A. Zadeh, “Knowledge Representation in Fuzzy Logic”, IEEE Transaction on Knowledge and Data Engineering, Vol. 1, No. 1, pp. 89-100, March 1989.
6. E. H. Mamdani, “Application of Fuzzy Logic to Approximate Reasoning Using Linguistic Synthesis”, IEEE Transaction on Computer, Vol. C-26, No. 12, pp. 1182-1191, December 1977.
SUMMARY OF THE INVENTION
This invention is a fuzzy logic controller based on the premise that at any given time all the input signal values to a controller must directly contribute to its output, each with an amount of influence on the output proportional to its level of strength. These strength levels can be described by certainty levels, with the highest certainty being 100% (or 1 on a scale of 0 to 1). It follows that each such input signal value must contribute to the fuzzy controller output with an amount proportional to its certainty level. This scheme is in agreement with the human operator decision which is based on the process situation which is, in turn, manifested in all the signal values received from the process. Furthermore, the scheme is also in agreement with the phenomena observed in dynamical systems that all the values of all the inputs to the system contribute to its output. In this way the fuzzy controller need only to fire a limited number of rules to achieve its task.
To accomplish the foregoing and other objects, and in accordance with the present invention as described and embodied herein, a system called the Multi Grade Fuzzy Logic Controller (MG-FLC) is presented embodying the organization, as well as, the scheme of implementation of a fuzzy logic controller system. In this system the measured process values are received as crisp data input and according to which control actions are generated in the form of crisp data output as control signals.
The following are the main aspects of the MG-FLC:
1. It is a controller for general purpose applications capable of solving linear and nonlinear control problems.
2. The controller can fire as few, or as many, fuzzy rules as desired, hence it can be configured to have fast real-time response.
3. There is reduced memory requirement.
4. The controller can solve complex and multivariable control applications.
In a preferred aspect, the present invention is directed to an apparatus for fuzzy logic inference engine comprising:
a) means for determining input membership function values by receiving signals of said input and determining the membership function value for every said input signal, such said input membership function values corresponding to an IF part of at least one inference rule;
b) means for determining an inference quantity corresponding to a THEN part of at least one inference rule, such said inference rule corresponding to said input membership function values;
c) means for determining an output signal by receiving all said determined input membership function values and every said inference quantity corresponding to a said THEN part of said inference rule corresponding to each said input membership function value; and
d) means for multiplying, adding, dividing and averaging to determine the said output signal values.
Other aspects and advantages of the invention will be apparent from the forthcoming description of the invention.
REFERENCES:
patent: 5485550 (1996-01-01), Dalton
patent: 5495574 (1996-02-01), Miyazawa et al.
patent: 5499319 (1996-03-01), Janabi et al.
patent: 5524179 (1996-06-01), Kanda
patent: 5572629 (1996-11-01), Choi
patent: 5787407 (1998-07-01), Viot
patent: 5796919 (1998-08-01), Kubica
patent: 5799132 (1998-08-01), Rizzotto
patent: 6041320 (2000-03-01), Qin et al.
patent: 6078911 (2000-06-01), Bonissone et al.
LandOfFree
Multi grade fuzzy logic controller does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Multi grade fuzzy logic controller, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Multi grade fuzzy logic controller will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2481149