D. Rakov, A.Timoshina (2010): Structure synthesis of prospective technical systems. IEEE Aerospace and Electronic Systems Magazine. - : Feb. 2010. - Volume: 25 -Issue: 2. - P. 4 - 10.

Structure synthesis of prospective technical systems.

pdf --> http://rakov.de/Artikel/rakov ieee systemanalyse 2010.pdf

**1.0. Introduction**

For creation of competitive products of aviation and rocket engineering it is necessary to synthesize from 50 up to 150 new engineering solutions.

In the work [8] 3 levels of optimization are examined at creation of new engineering solutions (ES). 1 level is understood as a choice of managing technical idea or a principle of action of projected system. 2 level of optimization is search of rational structure and 3 level - definition of the best values of characteristics for the chosen structure.

Performance characteristics of projected systems at the third level of optimization can be improved on the average on 10-15 %, and in some cases up to 30 %. Using 1 and 2 levels characteristics are improved on the average on 30-35 %, and sometimes in several times. The higher the level of optimization, the more effect of optimization is. In engineering practice usually there are no ways allowing at once to choose the optimum engineering solution based on conditions of the technical project . Therefore the process of development has iterative character. In the beginning the developer determines some array of alternatives which the projected system belongs to, and then tries to narrow this array, to test some engineering solutions, and to choose what is the most rational.

It is possible to divide the process of designing (fig. 1) into 2 stages. After statement of problem at 1 stage rational composition and the structure of system (qualitative characteristics) are chosen, and at 2 stage parametrical optimization with the fixed composition and structure (quantitative characteristics) is carried out. If necessary the process is repeated. Problems of parametrical optimization are given in to formalization and are well investigated. At the same time problems of 1 stage are difficult to formalize, and for their solution there is only small group of methods.

Fig. 1 The diagram of designing process

The problem is that the mistake in choice of ES can not be further corrected. Therefore it is necessary to analyse, as far as possible, all real variants which number can run up to several thousands. And accordingly, at the acceptance of the basic problem solutions there is a considerable excess of the information volume on potential variants of a choice in comparison with what the designer is able to process operatively [5].

**2.0. Morphological methods**

For structural synthesis the morphological method can be effectively used. It consists in construction of the morphological table, filling it by the possible alternative variants and in a choice from all array of the best solutions combinations. For the first time the method was applied by Swiss astronomer F.Tsvikki and henceforward it was developed in a number of researches [1,2,3,4,6].

Morphological methods suppose computer realization. The space of search is refered to as a morphological array, process of this space definition - the morphological analysis, and a search of the decision - morphological synthesis. As a result of the analysis there is a array of variants - alternatives. It contains all engineering solutions of an examined class devices, both real-life, and potentially possible.

The drawback of methods is the impossibility of the search and the analysis of all possible variants – potency of the morphological array can be enormous (to run up to tens and hundreds thousand possible alternatives).

**3.0. Structural synthesis of technical systems**

To reduce the dimensionality of the morphological array of a choice the method of the structural synthesis is developed. By means of the method it is possible to solve two groups of problems.** **

1. Direct problems. After the creation of the morphological array with the help of clusterization a search of engineering solutions occurs.

2. Inverse problems. The nearest vicinities in the searches of the more effective variants are looked through according to the known ES.

The synthesis process provides the following stages.

1.The creation of the morphological table.

The group of the basic characteristics is singled out in the object. Depending on a kind of a problem the essential characteristics {P}, i.e. able to affect the problem solution, are chosen from the array of characteristics{Šš}.

{Š} < = {Šš} (1)

The choice is the informal moment. For example, the array of characteristics {P} can be revealed from the formulas of inventions.

For each characteristic the elements are chosen, i.e. the possible variants of its execution or realization. Arranging them among themselves, it is possible to get the array of various solutions (variants).

The total number of variants equals:

N = A1 * A2 *...* Ai *... * An, (2)

Where Ai (i=1, n) - the number of values of 1,2... n the appropriate characteristic elements.

The morphological table (Fig. 2) contains 100800 potential variants.

The basic complexities on the way of choosing the solution are determined by two circumstances: complexity of formalization of a problem and the great amount of various requirements, criteria and restrictions. The criterion is a similarity of the purpose, its approximation, model. At the transition from the purposes to criteria, the last are considered as quantitative models of the qualitative purposes. The definition of the criterion value for the given alternative is, in essence, the indirect identification of its appropriateness as a means for the purpose achievement. Henceforth each element of the morphological table is compared with the appropriate value of the criterion on which the estimation will be made. The weighting coefficients are given to criteria depending on the purpose [7].

Fig. 2. Morphological table

2. At the second stage the generation of variants, their estimation, initial selection are carried out and some array of rational variants {R} for the subsequent analysis is formed. Each new generated variant is compared with the previous one from the array {R}. If it is at the higher level it is registered in the array {R}, if at the worse level it is rejected.

3. Henceforth the clusterization of the variants using the entered measure of similarity is carried out. The process of clusterization is considered as the search of a "natural" grouping of objects. The designer can choose the necessary degree of decomposition the initial array to clusters (fig.3).

Fig 3. A choice of the clusters quantity

The area of research is narrowed to several clusters (fig. 4) which are further investigated. Comparing variants, the best solutions which success is the most probable are being defined. The degree of the found alternatives novelty (fig. 5) is introduced.

Fig 4. Grouping the variants in clusters (6 groups)

Fig 5. The analysis of clusters and variants

4. After choosing some number of variants the parametrical optimization and the final choice are made.

**3.1. Inverse problems**

The process of searching a new engineering solutions is a subjective process. So designers have a psychological barrier – having found the first acceptable solution the process of search stops and the work with the chosen variant takes place. It’s intuitively obvious, that the first acceptable solution won’t be the best one, and better alternatives can be situated nearby. Psychologists have ascertained long ago, that in such situations people usually subconsciously single out just some variants of ES, and “forget” about the others [5].** **

For elimination of this drawback the method to solve the inverse problems of the structural synthesis [7,8] is developed.

The pivotal variant (idea) is being registered in the morphological array (fig. 6).

Fig 6. A morphological matrix with the pivotal solution

The further stages are the same as in the direct method. Those variants which measure of similarity is close to the pivotal are included in clusters (fig 7). The final matrix is made of the best alternatives. In a matrix there is an array of acceptable solutions on which the choice of the most suitable variant is made.

Fig 7. The process of the research of the pivotal variant vicinities

Various technical systems from the aerospace, power, ecological and medical areas were investigated by means of the method. For an example two engineering solutions - high-altitude stratospheric platforms and acoustic systems are given.** **

.....