Data processing: artificial intelligence – Neural network – Learning task
Reexamination Certificate
1998-04-06
2002-06-25
Davis, George B. (Department: 2121)
Data processing: artificial intelligence
Neural network
Learning task
C703S001000, C703S002000, C703S008000
Reexamination Certificate
active
06411945
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a design method, an optimization analyzing apparatus and a storage medium having a stored optimization analyzing program for a multi-component material and more particularly, to a design method, an optimization analyzing apparatus and a storage medium having a stored optimization analyzing program for design of a multi-component material composed of many components, for example design of a rubber compound for a tire.
BACKGROUND ART
Material design is to determine compositional ratios of components for a material to obtain the material having target mechanical behaviors, production conditions therefor and the like. The mechanical behaviors include physical quantities, such as physical properties of a material itself, sizes thereof and the like.
A conventional material design has mainly been conducted through experiences or trial and error and there has been difficulty in designing a material in the case of a material composed of three or more components. For example, in designing a rubber member for producing a tire, since various physical properties are considered, there have generally been taken procedures that a plurality of component materials having known physical properties are selected beforehand, the rubber member including predetermined compositions of respective component materials are as trial produced for tests and such tests are repeated until the member showing target performances in terms of a Young's modulus and the like is achieved. In this manner, the rubber member is designed and developed.
There has been proposed a material design method in which compositional ratio of multi-component materials can automatically be determined with use of a hierarchical neural network, if material properties are designated (see “Design Method for Multi-component Material Using Neural Network” in the papers presented at the first symposium on optimization sponsored by Japan Society of Mechanical Engineers, pp. 57-62).
In this method, however, if it is tried to determine compositional ratio of components by inputting material properties, which are slightly different from ones of an actual material, negative compositional ratio of the material is output, that is, compositional ratio of components with which a material cannot actually be composed is obtained and thus there arises a case where obtained compositional ratio cannot be used. Further, mechanical behaviors, such as material properties, except for mechanical behaviors which can be obtained the target performances cannot meet requirements from the market, so that the method has in fact been unable to apply in many practical aspects.
Another method has been known in which a polynominal is assumed for correlating compositional ratios of multi-component materials composed of a plurality of components and production conditions with mechanical behaviors of the materials and a calculating means determining coefficients of the polynominal by method of least squares and optimization are combined to each other (see authored by Manabu Iwasaki, “Planning and Analysis in Mixed Experiments” published by Scientist Co.,).
In this method, however, when optionality enters in assuming the polynominal or the number of the multi-component materials is three or more, the assumption of the polyniminal becomes difficult and the correlating cannot be performed with a high accuracy. For this reason, a plan of material design obtained has not been useful.
In view of the above facts, it is an object of the present invention to provide an optimization analyzing apparatus for a multi-component material and a design method therefor, in which a design of a material composed of a plurality of components is facilitated and since optimization is conducted while constraint conditions are considered, a design range of compositional ratios of components of multi-component materials and a desired range of a mechanical behavior can in advance be set.
DISCLOSURE OF THE INVENTION
The present inventor has studied various aspects in order to achieve the above mentioned object and as a result, has paid attention to application of a non-linear prediction technique, for example, a neural network, in which a neural circuit network of a higher animal is modeled in an engineering manner, and optimization design approach, both the technique and the approach being utilizing in some fields except for a material design field, to a special field of material design, carried out a research and established a design method in which mechanical behaviors, such as a Young's modulus, are considered in a concrete manner.
A design method for a multi-component material of the present invention comprises the steps of: (a) determining a conversion system in which a non-linear correspondence between compositional ratios of multi-component materials composed of a plurality of components and mechanical behaviors of the multi-component materials is established; (b) determining an objective function expressing the mechanical behaviors and setting a constraint condition constraining an allowable range of at least one of the mechanical behaviors and the compositional ratios of the multi-component materials; and (c) determining a compositional ratio of the multi-component materials which gives an optimal solution of the objective function on the basis of the objective function and the constraint condition by using the conversion system determined in the step (a) to design a multi-component material based upon the compositional ratio of the multi-component materials determined.
The mechanical behaviors of the multi-component materials composed of the plurality of components, such as Young's modulus, tan &dgr; and the like in a rubber member, are determined by the compositional ratios of components thereof. However, there are many cases where the mechanical behaviors are not changed in a linear manner, even though the compositional ratios are changed linearly. Therefore, in the step (a) of the present invention, the conversion system establishing a correspondence between the compositional ratios and the mechanical behaviors including a non-linear correspondence therebetween in advance is determined in advance. This conversion system can be determined with use of a non-linear prediction technique in which such a neural circuit network as neural network is modeled by an engineering approach.
In the step (b), the objective function expressing the mechanical behaviors is determined and the constraint condition constraining the allowable range of at least one of the mechanical behaviors and the compositional ratios of the multi-component materials is determined. As the objective function expressing mechanical behaviors, for example, a physical quantity governing excellency of a rubber member, such as a Young's modulus, tan &dgr; and the like can be used. As the constraint constraining the allowable range of at least one of the mechanical behaviors and the compositional ratio of the multi-component materials, there are in a rubber member, for example, constraints for a Young's modulus and Poison's ratio of the rubber member and a constraint for a mass of at least one component of the multi-component materials. The objective function, design variable and constraint condition are not limited to the above mentioned but various kinds can be determined according to a design object of a tire.
In the step (c), the compositional ratio of the multi-component materials which give an optimal solution of the objective function is determined based on the objective function and the constraint condition by using the conversion system determined in the step (a) and then the multi-component material is designed on the basis of the compositional ratio of the multi-component materials thus determined. In this manner, the conversion system is determined so as to establish the non-linear correspondence between the compositional ratios of components of the multi-component materials composed of a plurality of components and the mechanical b
Bridgestone Corporation
Davis George B.
LandOfFree
Method and apparatus for designing multi-component material,... 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 and apparatus for designing multi-component material,..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for designing multi-component material,... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2951725