Data processing: artificial intelligence – Fuzzy logic hardware
Reexamination Certificate
2000-09-14
2004-05-11
Starks, Jr., Wilbert L. (Department: 2121)
Data processing: artificial intelligence
Fuzzy logic hardware
C706S002000, C706S003000, C706S045000
Reexamination Certificate
active
06735576
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention relates to a method of optimizing a control device using fuzzy inference system so that a controlled system can be controlled.
2. Description of the Related Art
In the past, optimum values of the characteristics of a fuzzy controller (that is, values of the parameters such as the number, shape, position and/or expanse of membership functions, the fuzzy rules, and the standardized coefficients for input and output values) to control a controlled system were determined by experiment or experience in a design stage or a setting stage before shipment so that when use a environment and a user are supposed for a product that will be a controlled system, the supposed characteristics (preference, ability, personality, and use condition) of a use environment and a user can be met.
However, with the advent of diverse and highly advanced control technique, the conventional method of experimentally deciding optimum values for the characteristics of a fuzzy controller increases difficulty to optimize the fuzzy controller and needs to have a lot of time. Moreover, since a use environment of products, disturbance or a user's personality differs from individual to individual, the method does not enable all the users to satisfy the products of the characteristics.
SUMMARY OF THE INVENTION
To solve the problem, methods can be considered which, using fuzzy inference, neural networks, or heuristic rules, optimize output characteristics of a fuzzy controller that controls a controlled system in according to expected characteristics of a use environment and a user with a narrow range of variations which is also expected in advance. Since the methods can optimize the characteristics of the fuzzy controller in real time, the controller can handle variations of the characteristics of a use environment and a user.
However, the methods mentioned above have a problem in that when an variation of the characteristics of a use environment and a user, which cannot be expected when designing the system, takes place, the characteristics of the fuzzy controller cannot be optimized. The problem occurs not only in fuzzy controllers but also in controllers themselves having a fuzzy inference system. An embodiment of the present invention resolves the problem. Even when variations of characteristics of a use environment and a user occur, a control device having a fuzzy inference system can be optimized in real-time while the control device controls the controlled system, or the controlled system is being operated or on-line.
The method of real-time optimization of the control device in accordance with an embodiment of the invention (a) codes into chromosomes parameters used in a fuzzy inference system, which determines an output associated with a manipulated variable of a machine (controlled system) based on preselected input signals, and (b) optimizing in real time the output of the control device using evolutionary computation.
Evaluation of the evolutionary computation can be made based on evaluation criteria selected beforehand, or user's intention.
The parameters may be (i) the number, shape, position and/or expanse of membership functions for the fuzzy inference system of the control device, (ii) fuzzy rules, or (iii) standardized coefficients for input and output values.
The fuzzy rules can be compiled in the form of a fuzzy rule matrix. The configuration of the matrix may be defined by membership functions. Each section of the matrix represents a fuzzy rule which is a parameter having a value. The type of parameter and a value of the parameter are referred to as “a parameter”.
Coding into chromosomes can be made on all of the parameters or part thereof selected for the fuzzy controller.
Selection of part of the parameters can be carried out stochastically or deterministically starting from a membership function or a fuzzy rule having the highest total values of fitness in a fixed interval of time, which is the most frequently used parameters, until a preselected number of parameters are selected. To be specific, for example, a roulette selection of five positional parameters for membership functions are made proportional to the ratio of their total values of fitness to the overall fitness, or a selection of parameters for an expanse of membership functions is made in the order of size so that a total value of fitness can occupy over 80 percent of the entirety.
In selecting part of parameters, only some parameters that exceed a threshold of fitness fixed regarding the parameters can be selected. To be concrete, for example, depending on the ratio of a total of fitness to the entirety, a roulette selection of five fuzzy rules are made, or a selection of fuzzy rules are made in the order of size so that a total of fitness occupies over 80 percent of the entirety.
In deciding a threshold, the threshold can be made higher or lower depending on the numbers selected or the time spent till a selection is made.
As the fuzzy inference system, a minimax center-of-gravity method, an algebraic sum addition center-of-gravity method, a simplified inference method, a inference method having the degree of conviction, or a functional inference method can be employed. As the membership function, a one-dimensional membership function and/or a multi-dimensional membership function can be used.
As the evolutionary calculation, a genetic algorithm, an evolutionary strategy or an evolutionary programming may be used.
For purposes of summarizing the invention and the advantages achieved over the prior art, certain objects and advantages of the invention have been described above. Of course, it is to be understood that not necessarily all such objects or advantages may be achieved in accordance with any particular embodiment of the invention. Thus, for example, those skilled in the art will recognize that the invention may be embodied or carried out in a manner that achieves or optimizes one advantage or group of advantages as taught herein without necessarily achieving other objects or advantages as may be taught or suggested herein.
Further aspects, features and advantages of this invention will become apparent from the detailed description of the preferred embodiments which follow.
REFERENCES:
patent: 4785404 (1988-11-01), Sims et al.
patent: 4931025 (1990-06-01), Torigai et al.
patent: 5006992 (1991-04-01), Skeirik
patent: 5263432 (1993-11-01), Davis
patent: 5465204 (1995-11-01), Sekine et al.
patent: 5683275 (1997-11-01), Nanami
patent: 5813390 (1998-09-01), Anamoto
patent: 0 957 416 (1999-11-01), None
patent: WO 98/37465 (1998-08-01), None
patent: WO 98/50831 (1998-11-01), None
Janikow, Cezary Z., A Genetic Algorithm for Learning Fuzzy Controllers, Proceedings of the 1994 ACM symposium on Applied computing, Apr. 1994, pp. 232-236.*
Gopalan, Vijayarangan et al, Fuzzy genetic controllers for the autonomous rendezvous and docking problem, Proceedings of the 1995 ACM symposium on Applied computing, Feb. 1995, pp. 532-536.*
Pham, Trung T., Genetic algorithm for fuzzy modeling of robotic manipulators, Proceedings of the 1996 ACM symposium on Applied Computing, Feb. 1996, pp. 600-604.*
Chiu, Kuan-Shiu et al, GA design of crisp-fuzzy logic controllers, Proceedings of the 1999 ACM symposium on Applied computing, Feb. 1999, pp. 238-242.*
Ng, Kim Chwee et al, Design of sophisticated fuzzy logic controllers using genetic algorithms, Fuzzy Systems, 1994. IEEE World Congress on Computational Intelligence, Proceedings of the Third IEEE Conference on, Jun. 26-29, 1994, Page(s): 1708-1712 vol. 3.*
Tang, Kit-Sang et al, Minimal fuzzy memberships and rules using heirarchical genetic algorithms, Industrial Electronics, IEEE Transactions on, vol.: 45 Issue: 1, Feb. 1998, Page(s): 162-169.*
Chan, P.T. et al, An optimised fuzzy logic controller, Fuzzy Systems, 1997., Proceedings of the Sixth IEEE International Conference on, vol.: 2, Jul. 1-5, 1997, Page(s): 975-980 vol 2.*
D.D. Siljak, et al., On Hierarchic Optimal Control of Large-Scale Systems, 8thAsilomar Conference
Harada Hiroshi
Kaji Hirotaka
Matsushita Yukio
Yamaguchi Masashi
Knobbe Martens Olson & Bear LLP
Starks, Jr. Wilbert L.
Yamaha Hatsudoki Kabushiki Kaisha
LandOfFree
Method of real-time optimization of control device equipped... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of real-time optimization of control device equipped..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of real-time optimization of control device equipped... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3249887