Prediction of engineering solutions based on morphological approach

. Forecasting of scientific solutions based on the morphological approach, as well as the analysis and synthesis of new ideas and engineering solutions depends on a large number of different factors. This means that the specific results and duration of analysis and synthesis largely contain uncertainty factors. Uncertainty arises from the diversity of internal and external relationships, the complexity and variety of causal factors and the complexity with other engineering and social systems. The increasing need to increase the speed of adoption and development of new innovative engineering solutions determines the importance of building predictive models and finding methods to increase the compared alternatives and methods for their evaluation. The aim of the paper is to analyze the application of the improved morphological approach for predicting the development of complex systems. As a result, the analysis of the complex system of the system and forecasting of its development is carried out. The use of the proposed approach can be considered as a process of increasing the stability of synthesized systems and processes.


Introduction
In recent times, researchers have paid much attention to improving the reliability of scientific and technical prediction [1].The used forecasting methods, such as Delphi method [2,3], trend extrapolation [4], Integrated Assessment Models (IAM), scenarios [5], trend correlation and a number of others do not give full opportunity to take into account the factors of the external environment.The use of morphological approaches (MA) makes it possible to achieve the possibility of using various kinds of uncertainties in the models and taking risks into account.
The interaction of engineering system (ES) and subsystems is relatively simple and, in most cases, linear.Conflicts between subsystems are usually few and can be described.Therefore, the analysis and comparison of alternatives of engineering systems can be conducted as a cost-benefit or cost-effectiveness or cost-benefit analysis.In social systems the share of conflicts is significantly higher due to different interests and interpretations in the realization of such systems.Engineering efficiency is quite simply defined in terms of costs and effectiveness.Social efficiency as the opposite of technical efficiency is defined by a large number of mutually contradictory criteria and indicators.
When analyzing and synthesizing ES, it is advisable to be based on the provisions of system analysis.A number of researchers suggest studying the following regularities: - The forecasts themselves influence the decisions taken.The value of the forecasts can be measured by the degree of significance on the decisions made.Future facts cannot be predicted because they do not exist.The purpose of research in technical sciences is to construct and analyze alternative models.
Predictive models of the systems under study rely on the structure of the ESs under study.Quantitative analysis of ES is relatively simple in comparison with studies of social systems.
Fig. 1 shows statistics on published papers in the Scopus database with some prediction methods [6].

Morphological approaches
Classical methods of morphological analysis are based on the morphological box method of Fritz Zwicky [7,8].Morphological approaches can be used for a number of prediction cases [9].These methods are based on the creation of a morphological matrix (MM) (Figure 2).A morphological matrix consists of attributes (rows) and options (columns).The whole set of engineering solutions in the matrix constitutes the morphological set of solutions.Morphological approaches are widely used in various fields of science, engineering and economics [10][11][12].
The analysis is relatively often applied in the research of engineering systems and knowledge-based engineering (KBE-Knowledge-Based Engineering) [13].

Building predictive models
In the developed approach, the work is carried out according to the following algorithms (Fig. 3) [14,15]: problem formulation -MM creation with attributes and options, -assignment of a set of criteria, -evaluation of options, -assignment of reference options, -generation of solutions, -selection of some set of solutions, -construction of solution space using similarity measure and cluster analysis, -predictive analysis, -conclusions.The research was conducted with the help of the developed Okkam program [16].Experts use different criterion scores to build predictive models.Thus, in the example, 900 solutions are generated and 123 of them are selected with the best scores.Then they are grouped into 10 clusters.A system of 9 criteria is introduced for evaluation.7 criteria remain unchanged, while a group of environmental criteria receives different scores.Thus, in the first generation the scores of "Emissions" criterion is equal to 0.1 and in the second generation the scores are equal to 0.17.And correspondingly, in the first generation the scores of "Energy efficiency" criteria are equal to 0.17 and in the second generation the scores are equal to 0.27.Accordingly, the generated and selected solutions will be different.(Fig. 4-8)

Results
As a result of solution generation, the 2 clusters with the highest scores are selected.The number of solutions in the cluster of 10 in the first case was 5 solutions and 9 solutions in the second case (Fig. 9,10).The best solutions in the case of changing the values of the criteria will also be different.Figure 11 shows the generated and selected solutions.

Discussion
Forecasting of solutions based on the morphological approach, as well as the analysis and synthesis of new ideas and innovations depends on a large number of different factors.The analysis and synthesis of new solutions contains uncertainty factors to a large extent.This arises due to the diversity of internal and external relationships, as well as interaction with other engineering and social systems.The paper is devoted to analyzing the application of an improved morphological approach for predicting the development of systems.As a result, a complex system is analyzed and its development is predicted, and the best solutions are synthesized.The use of the approach is to increase the stability of the synthesized systems.Consequently, the probability of synthesizing stable engineering solutions increases.
Interaction of techniques and technologies at different stages; -Study of the change of generations of techniques and technologies; -Study of diffusion of high and critical technologies; -Development of technologies to create new ES; -Conducting scientific research at all stages of R&D; -Use of artificial intelligence tools in production and design processes; -Use of structural optimization of advanced engineering systems and technological processes; -Management of R&D projects and execution of experimental and technological works Management of innovation projects.

Fig. 2 .
Fig.2.A morphological matrix of attributes and options (left) and an example of a real matrix (right)