ПІДХІД НЕЧІТКОЇ КЛАСТЕРИЗАЦІЇ В РОЗПОДІЛЕНИХ ІНФОРМАЦІЙНИХ СИСТЕМАХ АВІАЦІЙНИХ ДВИГУНІВ

Abstract

The article is devoted to the development of systematic approaches for the construction of a distributed information system (DIS) of aviation gas turbine engines (GTE). It is determined that the use of CALS-technologies (Continuous Acquisition and Life-Cycle Support), which should ensure the competitiveness of products on the world market, is essential for the integration of the aircraft engine industry into the world community of developers and manufacturers. The relevance of the use of CALS-technologies is due to the fact that today, in accordance with market requirements, the world's leading companies have set deadlines for the creation of a new design of the civil aviation engine of the fifth and sixth generations. The block-modular principle of engine construction - mathematical models and software - with the satisfaction of the criteria of divergence, transformation, and convergence has been considered. For a simplified search of the optimal technology for building a distributed information system of an aviation engine, the use of a fuzzy clustering approach is proposed, which is a design method with finding new knowledge about the gas turbine engine with highly efficient performance. By identifying methods of knowledge analysis and basic methods of clustering, that are K-means, graph clustering algorithms, algorithms of the FOREL family, hierarchical clustering, Kohonen neural network, algorithms of the KRAB family, fuzzy mean algorithms, subtractive, the application clustering in distributed information systems of aviation engines have been determined. For the convenient implementation of the defined method, a set of data objects of the GTE information system, which are contained in the experimental files, are considered. According to the results of the fuzzy clustering procedure, the coordinates of the class centers, the belonging of each data set to the classes, the values of the objective function, which have an approximate character, and are used for preliminary structuring of the data, are fixed. After research, it was determined that the integration of clustering algorithms should help build a more accurate model of the gas turbine engine and increase its speed.