Technical science/12. Automation control systems in manufacturing

Ph.D., Salykova O.S.

Undergraduate of specialty 6M070400- Computing and software, Bizhanova O.I.

Akhmet Baitursynov Kostanay State University, Êàçàõñòàí

PROBLEM DEFINITION OF OPTIMIZATION AND METHODS OF ITS DECISION FOR DESIGN OF THE AUTOMATED SYSTEM

 

Investigating this or that task of optimization, first of all it is necessary to choose a mathematical method for its decision. The chosen method has to lead to such results that the volume of information was the greatest, and costs of calculations – the smallest. The main criterion of a choice of a method is not only statement of an optimum task, but also also mathematical model which use in the course of object optimization.

The methods solving optimum problems a set, but the following is the most popular:

-       methods of research of functions of the classical analysis;

-       the methods based on use of uncertain multipliers of Lagrange;

-       calculus of variations;

-       dynamic programming;

-       principle of a maximum;

-       linear programming;

-       nonlinear programming.

Since recent time the method of geometrical programming was developed and introduced. This method use for the solution of a certain class of tasks.

As a rule, it is difficult to recommend only one method which can be used for the solution of all tasks. One methods are more the general, others – less the general. The whole group of methods (methods of research of functions of the classical analysis, a method of multipliers of Lagrange, methods of nonlinear programming) at certain stages of the solution of an optimum task can be applied in combination with other methods, for example dynamic programming or the principle of a maximum.

Dynamic programming is well adapted for the solution of problems of optimization of multistage processes, especially in what the condition of each stage is characterized by rather small number of variables of a state. However in the presence of considerable number of these variables, i.e. at high dimension of each stage, application of a method of dynamic programming is difficult owing to the limited speed and memory size of computers.

Dynamic programming serves as an effective method of the solution of problems of optimization of discrete multistage processes for which the criterion of an optimality is set as additive function of criteria of an optimality of separate stages. Without special difficulties the specified method can be extended and to a case when the criterion of an optimality is set in other form, however thus dimension of separate stages usually increases.

In essence the method of dynamic programming represents algorithm of definition of optimum strategy of management at all stages of process. Thus the management law at each stage find a solution of private problems of optimization consistently for all stages of process by means of methods of research of functions of the classical analysis or methods of nonlinear programming. Results of the decision can't be usually expressed in an analytical form, and turn out in the form of tables. Restrictions on variable tasks have no impact on the general algorithm of the decision, and are considered at the solution of private problems of optimization at each stage of process. In the presence of restrictions like equalities sometimes even it is possible to reduce dimension of these private tasks due to use of multipliers of Lagrange. Application of a method of dynamic programming for optimization of processes with the distributed parameters or in problems of dynamic optimization leads to the solution of the differential equations in private derivatives. Instead of the solution of such equations often it is much simpler to present continuous process as discrete with rather large number of stages. Similar reception is justified especially when there are restrictions on variable tasks and the direct solution of the differential equations becomes complicated need of the accounting of the specified restrictions.

As the best way at a choice of a method of the optimization which is most suitable for the solution of the corresponding task, it is necessary to recognize research of opportunities and experience of application of various methods of optimization.

 

Literature:

1.     1 . A. G. Trifonov. "Problem definition of optimization and numerical methods of its decision", 2012.

2.     2 . The automated design of technical systems: Manual. Baby seals V.N., Lanshakov V. L. Natural Sciences Academy publishing house, 2009, ISBN 978-5-91327-056-6.