Data processing: artificial intelligence – Machine learning – Genetic algorithm and genetic programming system
Reexamination Certificate
1999-05-24
2001-11-27
Powell, Mark R. (Department: 2122)
Data processing: artificial intelligence
Machine learning
Genetic algorithm and genetic programming system
Reexamination Certificate
active
06324529
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to an evolutionary controlling system, and particularly to that for controlling characteristics of a subject in an evolutionary manner.
2. Description of Related Art
Heretofore, when a control system or control characteristics of a subject, such as vehicles and electrical appliances, is designed, imaginary users are selected, and the users' preferences and their using conditions are taken into consideration. The characteristics of the subject are determined in such a way as to adapt the subject to users in as broad a range as possible.
However, each individual user has a particular and unique personality, and thus, their preferences are diverse. Thus, there is a problem in that even if imaginary users are selected to develop and design a product for users by presuming the users' preference, it is impossible to satisfy all of the users of the product.
In order to solve the above problem, prior to purchase of a product, a prospective user is requested to determine whether or not the product is satisfactory to the user after checking the characteristics of the product in light of the user's preferences. However, it is troublesome for the user to check the characteristics of the product before purchase. Further, because a series of products are often operated or controlled by characteristics common in the products, although the design of the product is changed depending on the user's preferences, the user may not like other operational characteristics. Thus, although the design is appealing to some prospective users, the users may not purchase the product since the operational characteristics do not appeal to them. In other words, there is another problem in that the range of users is limited and depends on the operational characteristics.
An objective of the present invention is to provide an evolutionary control system to construct characteristics which can satisfy plural users.
SUMMARY OF THE INVENTION
One important aspect of the present invention includes a method for controlling performance of a machine used by a user, which machine is operable by causative signals, the performance of which machine is indicatable by indicative signals. In the above, the machine is controlled by a control system comprising: (I) a control module programmed to output the causative signals when receiving per-selected signals, wherein the input-output relationship of the control module is regulated by coefficients; and (II) a genetic algorithm unit programmed to select fitted coefficients based a selection signal when using coefficients as genes, wherein the selected fitted coefficients replace the coefficients used in the control module to update its input-output relationship. One embodiment of the method of the present invention comprises the steps of: (a) outputting the causative signals from the control module to the machine when the control module receives per-selected signals; (b) selecting a selection signal from the group consisting of: (i) a selection signal inputted directly by the user in response to the performance of the machine based on its indicative signals; (ii) a selection signal expressed numerically by analyzing the user's reaction to the performance of the machine based on its indicative signals; and (iii) a pre-set target value; (c) selecting fitted coefficients by the genetic algorithm unit based on the selected selection signal; and (d) replacing the coefficients used in the control module to update its input-output relationship to control the machine. One practical example can be found in a vehicle engine wherein fuel efficiency can be improved autonomously (by using the pre-set target value as a selection signal) whereas drivability can be improved in accordance with the user's response (by using the direct selection signal or the user response signal), thereby improving integrally the performance of the engine.
FIG. 1
is a schematic diagram illustrating embodiments of the method including the above.
As shown in
FIG. 1
, the method can further comprise, prior to step (a), inputting the causative signals from the control module to a computer simulation model by bypassing the machine, said simulation model programmed to simulate the performance of the machine, wherein steps (a) through (d) are conducted using the simulation model. The control module may include a neural network, the input-output relationship of which is regulated by coupling coefficients, said coupling coefficients being used as genes at the genetic algorithm units. In another embodiment, the control module constitutes an evolution layer, and the control system further comprises a base layer downstream of the evolution layer and upstream of the machine. In the above, the method further comprises calculating and adding, by the base layer, base values of the causative signals to the outputs from the control modules of the evolution layer, based on pre-selected signals. The base layer is not indispensable, but advantageous in preventing surges of the causative signal and stabilizing the performance. The base layer can be a conventional control module which constantly outputs signals based on calculation.
In anther embodiment, the method further comprises a learning layer between the evolution layer and the base layer, said learning layer having a learning function which copies the input-output relationship of the control module. In the above, the method further comprises copying the input-output relationship of the control module and outputting causative signals to the base layer. The learning layer can be constituted by, for example, neural networks.
In another embodiment, the user's direct selection is conducted by using an indication monitor which indicates the performance of the machine, wherein the user selects preferable performance indicated on the monitor. Further, in another embodiment, the user reaction is analyzed based on a change in operational action by the user in a pre-set time period.
The present invention can be applied not only to a method but also to a system. An appropriate system can be constructed accordingly. In addition, although the present invention can advantageously and preferably be applied to an engine, it can be applied to other machines as described above.
REFERENCES:
patent: 5701400 (1997-12-01), Amado
patent: 5761384 (1998-06-01), Inoue
patent: 5819242 (1998-10-01), Matsuoka et al.
patent: 5841949 (1999-11-01), Nakaya et al.
patent: 6032139 (2000-02-01), Yamaguchi et al.
patent: 6064996 (2000-05-01), Yamaguchi et al.
patent: 6216083 (2001-04-01), Ulyanov et al.
patent: 0959414A1 (1999-11-01), None
patent: 0962871A2 (1999-12-01), None
Apple Advanced Technology Group, Vivarium Program, Artificial Life II Conference Report by Larry Yaeger, (Feb. 5th through 9th, 1990) Sweeney Center, Santa Fe, New Mexico.*
Artificial Evolution: A Neww Path for Artificial Intelligence? P. Husbands, I. Harvey, D. Cliff, and G. Miller (Brain and Cognition 34, 130-159 (1997) Article No. BR970910).*
The Hitch-Hiker's Guide to Evolujtionary Computation (FAQ for comp.ai.genetic) by Jorg Heitkotter and David Beasley (1997).*
The design of natural and artificial adaptive systems, Frank, S. A. (1996) pp. 451-505, in Adaption, M. R. Rose and G. V. Lauder, eds. Academic Press, New York.
Kamihira Ichikai
Yamaguchi Masashi
Holmes Michael B.
Knobbe Martens Olson & Bear LLP
Powell Mark R.
Yamaha Hatsudoki Kabushiki Kaisha
LandOfFree
Evolutionary controlling system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Evolutionary controlling system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Evolutionary controlling system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2592514