• DonNTU
• Master's portal

• Русский
• Українська
• English
• Deutsch

• Biography
• Master work

Abstract of Master work
MIMD-simulator and optimizer parallel models of discrete dynamic systems

Introduction

Simulation as one of the most important categories of knowledge can not be separated from human development. A man uses a model that can play any life and work situations and to obtain such solutions, which allow to find the best way to solve the problem. In such cases it is possible to analyze with the help of the model, any situation, including those in which the real system would be left out. This makes it possible to simulate disaster rare cases, even such phenomena and processes that do not exist in reality, it is virtual reality [1].

Methods of computer simulation is widely used in all spheres of human activity - from the construction of models of technical, technological and organizational systems to address the problems of mankind and the universe. Classical objects are information modeling, manufacturing, transportation and other logistics systems. Very promising is the use of models in the control loops, i.e., in real time. The most important task of simulation is to evaluate the performance of such systems.

Urgency

Today has been a rapidly evolving situation in the market, exacerbating the economic situation. Large enterprises in a competitive environment with the help of simulation and optimization are able to react quickly to market changes. They can plan their actions and develop a strategy for using the model as well as to avoid possible mistakes and financial losses. This helps to strengthen the competitiveness of enterprises.

Companies turn to automation, decentralization, and to create virtual networks of production areas. All this complicates the production process as a whole, as well as the processes of logistics. Simulation and optimization are helping to support the development of these processes at the appropriate level.

In planning the production process needed to solve such problems as the division of work and time, the optimal use of resources, to develop a transport system, and others. All of this ultimately determines the effectiveness of the strategy of the enterprise. Therefore, in order to guarantee optimal solutions in processing such a large amount of information that would be appropriate to use computer simulation and discrete optimization of existing systems.

The purpose and objectives of development and research

The aim of this work is to develop new approaches to the parallelization and optimization models of discrete dynamical systems. To do this task: develop a test Queueing model as a kind of discrete dynamical systems, with the help of discrete simulation and embedded system optimization ISSOP / Arena.

Supposed scientific novelty

Will develop new approaches to the parallelization and optimization models of discrete dynamical systems, using existing sequential software.

Planned practical novelty

Development of test models of queuing, as well as the development of test methods for constructing discrete models using discrete simulation ISSOP / Arena.

The main content of work

Dynamic objects are called engineering and technology, in which targeted the processes of change of the state-administered, accompanied by a change in parameters over time. Dynamic systems characterized by two fundamental mode operation: a stationary regime with constant parameters and the transition (dynamic) mode with the change of parameters.

DDS is a system with the discrete nature of dynamic processes. Common to the investigated DDS is the presence of various random factors that significantly influence the change of states. It is assumed that the set of states of the investigated system is discrete and the change of states occurs at some points in time. The intervals between the moments of change of states can be as random or deterministic values. During the interval between the moments of change of conditions studied the system state does not change.

The Queueing model (QM) serves claims that are coming into it from the source of claims and returns after service in the spring. Service requirements in the system is performed by service technicians. The system may contain from one to the infinite number of devices [2]. Depending on the implementation of the system the possibility of waiting received their maintenance requirements of public service are divided into three types:

1) of losses in which the requirements are not found at the time of receipt of any free device is lost (come back to the source without the service);

2) system with the expectation that perhaps the expectation of any number of claims, while claims are pending queue length is not limited;

3) system with the expectations and limitations, which allowed the formation of a queue of limited length, with the demands made in the system where there are no free places for waiting in the queue, get lost (come back to the source without the service).

In systems with a queue waiting in the general case may have a complicated structure, a certain set of queues. The choice of the next requirement in the queue for services performed through a discipline of service. Examples of such disciplines are FCFS (first come - first served), LCFS (last come - first served), RANDOM (another requirement is chosen at random from the queue). A random sequence of claims, which come in care and to be serviced, is one of the fundamental concepts of queuing theory. These sequences are also known as flow requirements.

Simulation is a method that allows to build models that describe the processes as they would take place in reality [3]. Such a model can be «play» in time for one test, and given their set. The results will be determined by random processes. According to these data can be obtained fairly stable statistics.

Simulation of - this method of research, which studied the system with sufficient accuracy be replaced by a model describing the real system and its experiments with the purpose of obtaining information about the system. Experimentation with the model referred to as imitation (simulation - this is indeed understand a phenomenon without resorting to experiments on a real object).

Simulation is a special case of mathematical modeling. There is a class of objects, which for various reasons, there are no analytical models or have not developed methods for solving the obtained models. In this case, the mathematical model is replaced by simulator or simulation model.

Simulation Model - logical-mathematical description of the object, which can be used to experiment on your computer in order to design, analyze and evaluate the operation of the facility.

One type of simulation - discrete Events Mod. This approach to modeling, offers apart from the continuous nature of the events and consider only the major events of the modeled system, such as: «wait», «Customer», «movement control», «unloading» and others. Discrete-Events Simulation most developed and has tremendous scope of applications - from logistics and systems services to the mass transport and production systems. This type of model best suited for the simulation of manufacturing processes. Geoffrey Gordon was founded in 1960s.

The main value of simulation is that it is based on a methodology for system analysis [4]. It provides an opportunity to examine or test a system designed along the lines of operational research, including the interrelated stages: the content of the problem, the development of a conceptual model, design and implementation of simulation model checking the adequacy of the model and estimate the accuracy of the simulation results, the planning of experiments, decision-making. This simulation can be used as a universal approach to decision making under uncertainty and uchityvaniya in models of the factors that heavily formalized, as well as in practice to use the basic principles of a systematic approach for solving practical tasks. However, the implementation of simulation models and conducting experiments, it is laborious, expensive. This requires a considerable amount of computer time, so simulation requires contact only when the development of other types of models does not give satisfactory results.

The software for modeling appeared in the early 50's [5]. To date, distinguish 5 major generations modeling tools:

1. The first generation there was a long time, accompanied by a large number of errors, and required a specialist in modeling and informatics.

In the second phase appeared in simple language modeling, such as GPSS, who gave the opportunity to do a statistical analysis of results.

3. In the third phase, there wide range of combinations of modeling languages such as GPSS and FORTRAN.

4. The fourth generation as a decisive criterion for success has been able to display on-screen work model.

5. At the fifth stage of modeling is aimed not only to simulate the process, as well as the optimization done through a standard open interface.

The simulation system of 5-th generation is ARENA.

GPSS language was the language that defined the modern technological trends in the discrete simulation and was the forerunner of modern languages and systems for modeling of discrete-type [6]. These trends are determined primarily successfully established the basic scheme of the structure, built in GPSS, supports block-oriented approach, in which the modeling unit has a functional purpose and submitted to the relevant functional objects (having similar to elements of public service), as well as opportunities for language description of parallel processes. This is the view of the simulated object is allowed to realize ideographic mode of forming the discrete model when the model is constructed of standard functional blocks, and connection to these graphic designs are interpreted as the route of the moving objects in the system.

Disadvantages:

Usually observed deficiencies GPSS (integer model time, the interface is weak, poor visualization of the modeling process and its outcomes, the complexity of integration with other programs and databases, weak support for mathematical functions, lack of flexibility, etc.) were readily retrievable in a new technological level with the use of flexible language.

Basis Technology Arena - SIMAN simulation language and system of Cinema Animation [7]. SIMAN, first implemented in 1982. - Extremely flexible and expressive modeling language. He is constantly being improved by the addition of new features. To display the simulation results using animation system Cinema animation, known on the market since 1984, the process simulation is organized as follows. First, the user step by step in building a visual editor of Arena model. The system will then generate the appropriate code for it to SIMAN, then automatically starts Cinema animation.

Interface Arena includes various tools for working with data, including spreadsheets, databases, ODBC, OLE, supported formats DXF.

Integrated modeling and optimization ISSOP / ARENA:

ISSOP / ARENA has been developed based on experience in modeling and optimization, it is able to find complex solutions to the problems of optimization with the help of powerful mathematical strategies. The concept of PSE

Parallel simulation environment (PSE) - a set of parallel hardware, as well as tools and models for systems that support all stages of development, debugging, and application models.

The structure of the PSE

Fig. 1 Structure of the PSE (animation, 13 frames, 5 cycles, 114 KB)

As the hardware for the PSE can be used concentrated parallel computers (SIMD, MIMD) and networked computer systems (CLUSTER, META).)

As the system software for computers must be focused operating system and parallel programming languages for networked computing systems is the network operating systems and networking software.

The basis of the simulator software must be a system ISSOP and Arena.

Conclusions

With the successful development of test models of queuing systems and methods for constructing discrete test models using discrete simulation ISSOP / Arena, you can use the results obtained in practice for the development of students.

References

1. Томашевсский В.М. Моделювання систем. - К.: Видавнича група BHV, 2005 - c. 9-14

2. Система массового обслуживания [Электронный ресурс] http://ru.wikipedia.org/wiki/Система_массового_обслуживания

3. Имитационное моделирование [Электронный ресурс] http://ru.wikipedia.org/wiki/Имитационное_моделирование

4. Томашевський В.М. Iмiтацiйне моделювання систем i процесів: Навч. посібник. - К.: IСДО, 1994. - 124 c.

5. Krug W. Intelligentes Simulations - und Optimierungsystem fur Prozesse der Fertigung, Organisation und Logistik ARENA/ISSOP. - Ghent.: SCS-EUROPE BVBA, 1997. - pp. 1-4

6. Лычкина Н. Опыт применения GPSS в Государственном Университете Управления [Электронный ресурс]
http://www.gpss.ru/paper/lychkina/1_w.html

7. Антипина Г., Ярцев А. Arena - система имитационного моделирования [Электронный ресурс] http://www.interface.ru/sysmod/arena.htm

8. Krug W. Modelling, Simulation and Optimisation. - Erlangen.: Gruner Druck GmbH, 2002. - 267 p.

9. Бусленко Н.П. Автоматизация имитационного моделирования сложных систем. М.: Наука, 1977. - 240с.

10. Бусленко Н.П. Моделирование сложных систем. М.: Наука, 1978, 400с.

11. Шрайбер Т.Дж. Моделирование на GPSS: М.: Машиностроение, 1980. - 583 c.