Ontological metamodeling of grain drying processes · 2019. 6. 24. · S. Arhipov Latvia University of Agriculture, Liela street 2, LV – 3001, Jelgava, Latvia, [email protected]

EFITA/WCCA 2005 25-28 July 2005, Vila Real, Portugal

2005 EFITA/WCCA JOINT CONGRESS ON IT IN AGRICULTURE

Ontological metamodeling of grain drying processes

S. Arhipov

Latvia University of Agriculture, Liela street 2, LV – 3001, Jelgava, Latvia, [email protected]

Abstract

The present paper deals with the traditional architecture of modeling, supplemented with two new approaches from the linguistic and ontological points of view. These two schemes of metamodeling are exemplified with the conceptions from the grain drying theory. The application of different forms of concretization and abstraction enables to mark off the metalevels in two orthogonal systems. This increases the abstraction level when the sophisticated problems are solved, that might change crucially the approaches applied to the system modeling.

Key words: Metamodeling, ontology, grain drying.

1 Introduction

Preliminary formation of specialized models helps to solve the sophisticated problems and to analyze the techniques to deal with these problems by means of abstract conceptions. Therefore application of models and modeling technologies ensures significant advantages when designing sophisticated classes of engineering systems (J.Mukerji, J.Miller, (Eds.), 2001).

Model is a presentment of one or several objects, which has one of the qualities (Ian Graham, 2001). • This presentment differs from the objects-originals to be modeled by scale, realization and behavior. • Model has the form characteristic to the original. • Model is used to forecast the behavior and qualities of the original (simulation model). • There is a certain correspondence between model and original.

The model reflects the separate qualities of the reality to be modeled. Insignificant details in the model are simplified or abandoned. The model is formed in a particular environment by means of given rules and axioms. For example, the model of grain drying might be reflected as a graph of temperature, humidity and mass. But the system of mass transfer and thermal conductivity equation reflects the mathematical model of drying. The model of the system is formed by means of the modeling language, for instance, Unified Modeling Language (UML) (Jon Siegel, (Eds.), 2001).

2 Traditional infrastructure of modeling

Unified Modeling Language is based on the four-level hierarchy of models that is shown in Figure 1. Every level of hierarchy is an abstraction, the degree of which increases from lower to upper level.

The lowest level comprises the actual objects of the given object range. The object notion of the object range is the general conception and it does not have formal definition. The total of the objects that form the core of the object range has the ontological status. The term “object” is initial and does not determine. The synonyms of the object notion are “reality”, “thing” and “entity”. The samples of objects in the drying theory are body, humidity, heat, drying agent, grain-dryer, separate grain, grain mass, drying process.

172



Fig.1 Traditional structure of modeling

The second abstraction level determines the model of the object range that is stated in the lower abstraction level. If the first abstraction level indicates objects as particular entities, the second abstraction level determines common structure and behavior of the certain quantity of objects. Any specific object is an instant of class. Thus we can speak about “Grain” class that comprises qualities, which are common to all “theGrain”. The passage from the first abstraction level to the second one models the nature of the relation type “object is a class instance”.

However, particular relation “A is B” can be used also regarding the classes: the classes are also metaobjects above the classes or instances of metaclasses. This modeling standard is implemented on the third abstraction level. At this level the information from the second level is modeled. In this case we speak about “the model of the model”, therefore the third level is called the metamodel.

The fourth level contains the model of the metamodel and it is call the meta-metamodel or Meta-Object Facility (MOF).

The infrastructure of this modeling envisages that the basic nature of the relation type “object A is an instance of class B” is constant. However, the simplified reflection of the formation of an instance can be viewed from two orthogonal positions – linguistic and ontological.

3 Linguistic metamodeling

The linguistic analysis of the object range enables to identify the real objects and to model the object classes. The model of classes in Figure 2 reflects the types of the object range: The humid materials are the capillary porous colloidal bodies. The capillary porous colloidal bodies belong to the class of coherently disperse systems, where the particles of dispersion phase form more or less fixed spatial structures – lattices or frames. These systems are called gels. All humid materials are divided into three types: elastic gels, fragile bodies and capillary porous colloidal bodies, which have the qualities of the first two.

Meta-models

Model of UML

User model

User dates

173



Fig.2 Object range classes

Within the context of linguistic modeling the relation “A is an instance of B” is regarded as prevailing, but the ontological relation “A is an instance of B” has a minor role. The linguistic relations form the abstraction levels, but the ontological relations attribute objects to each other within the particular level. Figure 3 shows the linguistic interpretation of the four-level architecture that is performed within the standards of UML and MOF.

The real grain is shown as the object theGrain, which is an instance of the class Grain. User’s objects and the classes that form them ontologically are situated on the same linguistic level. Every level is a linguistic model that is expressed in the language determined in the above situated level.

4 Ontological metamodeling

The linguistic metamodeling does not enable to extend dynamically the types and metatypes of the object range. The description of such conceptions and their qualities is called the ontological metamodeling.

Figure 4 shows an example of the ontological formation of instances. The element of the model theGrain is the ontological instance of FirmWheat. The idea FirmWheat is the logical type of theGrain in reality. In its turn, FirmWheat is the ontological formation of the metatype Wheat, which is situated one level above. The ontological metatype enables not only to differ one type from another, but also to input the new qualities that are characteristic to all formed types.

Fig.3. Presentment of the linguistic metamodelling.

174



Ont

olog

ic

Inst

ance

On t

olo g

ic

Ins t

ance

Ont

olog

ic

Inst

ance

On t

olog

ic

I ns t

a nc e

Fig.4. Presentment of the ontological metamodelling.

For example, the variety can differ such types as hard wheat and soft wheat. We can also use the type “variety” to determine the qualities of the variety – growing geography and the genealogical links with other varieties.

Metaconceptions “variety”, “species”, “family” enable to extend the systematic with the new classes of crops. In the program system they would facilitate the dynamic addition of new types in the process of performance.

5 Ontology of grain drying

The phenomenon of the heat and humidity transfer is described by means of the diffusion model. The system of particles that forms the substance is indeterminate. We look at the system that consists of particles, which move chaotically. The system is indeterminate in a sense that, knowing the initial state of a particle in a space, we are not allowed indicate directly its coordinates at any coming moment. It is the class of the particle movement tasks, where the particles implement their movement under the influence of incidental impulses. For instance, the movement of the diffusive substance molecules is performed under the influence of the incidental shocks of molecular environment. It is impossible to anticipate the movement of every particle, but only to forcast with a certain probability. However, it is excellent that, if the particles, which move chaotically and independently from each other, are in great number, it is possible to determine unequivocally the behaviour of the particle system in general.

Two types of environment take part in the diffusion process. One of them diffuse, but the other one serves as a base, where the diffusion takes place. Such division does not have the complete characteristic: it is possible, when the concentration of the diffusive environment is low.

+Agent()+~Agent()+Move(in r : Random) : int+Draw(in g : Graphics)

WATOR::Agent+Temperature : float+Humidity : float+Adress : Point

+Move(in r : Random) : int+Draw(in g : Graphics)+CoverAgent()

WATOR::CoverAgent

+Draw(in g : Graphics)+GrainAgent()+Move()

WATOR::GrainAgent

+Draw(in g : Graphics)+Move()

WATOR::DryAgent

Speed+x : int+y : int

1

+speed

1

Fig.5. Ontology of grain drying process.

175



CenterLef Right

Top

Bottom

[p=0.2]

[p=0.2]

[p=0.2]

[p=0.2]

[p=0.2]

/ Start

/ Stop

/ Right = Center

/ Top = Center

/ Left = Center

/ Bottom = Center

Fig.6. State diagram of moving of agents.

During drying the structure of grain does not change. Grain only loses humidity. But the skeleton of grain imposes restrictions on movement of diffusive humidity. Therefore in static model three classes are considered. It is a class, which models drying agents, a class, which models agents of structure of grain, a class, which models agents of moisture. These three classes form a basis of grain drying ontology (figure 5).

Three classes have the common properties. It is a temperature and humidity. But they are differed by behaviour. Really, drying agents and agents of moisture are capable to move, but structural agents of grain remain motionless.

It is meaningful to enter into models of drying linguistic concept of the Agent which generalizes ontological agents. Class the Agent realizes the common for all agents of system properties "temperature" and "humidity" and models their generalized behaviour. In this case the behaviour of the generalized Agent is modeled by means of state diagram (Figure 6).

Each concrete class of the agent specifies this behaviour. Parameters of movement of the drying agent depend on value of temperature. Parameters of movement of the moisture agent depend on value of humidity. Grain structural agents are motionless.

6 Modelling of drying process

The diffusive particles are concentrated in the middle of line segment. In Figure 7 the high concentration of particles is showed in black. Where the concentration is low or equal to zero, it is showed in white. The black color indicates the substance concentration between the high and low value. Every particle moves into one of its two neighbouring points or stays where it is.

Fig.7. Diffusion process of drying agent

176



Fig.8. Interaction of the diffusion process of particles with the coat

Fig.9. The diffusion of the particles from the body into the environment

In next experiments (figure 8) time axis is showed horizontally. There were performed about 1000 iterations. Can see that in the course of time the particles become as a cloud with dim sides. The drying agent tend to be positioned along all the length of line segment, but at the same time we can see that the concentration of particles is higher in the place, where they were generated initially.

There was repeated the experiment (figure 9) with the diffusion of the drying agents. The heat is inside the body, where there is a hole. The body coat is showed in grey. The body coat consists of the particles, which cannot move and which are the obstacle for the other particles. The black colour indicates the high concentration of particles. The experiment shows how decreases the concentration of squares inside the body and increases their concentration in the environment. This process can be interpreted as the transfer of humidity and heat from the inside of body into the environment.

The experiment (figure 9) shows the location of the immobile body in the diffusive flow of particles. The particles to be modeled are able to escape obstacles. We can observe the high concentration of particles in the lower part of body and the low concentration – in the upper part of body. This experiment shows the modeling of limits within the diffusion process.

7 Conclusions

Linguistic and ontological metamodeling have equal power and they are equally useful, because they do not differ within the traditional modeling structure. Both the schemes supplement each other – when the linguistic approach is applied, the secondary role is given to the ontological metamodeling. The linguistic modeling backs away, if the ontological point of view dominates.

8 References

J.Mukerji, J.Miller, (Eds.), July 9, 2001. Model Driven Architecture (MDA), Document number ormsc/2001-07-01, Architecture Board ORMSC1.

Jon Siegel, (Eds.), November, 2001. Developing in OMG’s Model-Driven Architecture, Object Management Group, White Paper, , Revision 2.6

Ian Graham, 2001. Object – Oriented Methods, Principles and Practice, 3rd Edition, Addison-Wesley.

177

Ontological metamodeling of grain drying processes · 2019. 6. 24. · S. Arhipov Latvia University of Agriculture, Liela street 2, LV – 3001, Jelgava, Latvia, [email protected]

Documents