Özet:

Abstract The problem of automation synthesis of neuro-fuzzy networks for diagnostics and automatic classification on features is solved. The method of neuro-fuzzy model synthesis on precedents is proposed. It evaluates the relationship of input features, extracts related features, evaluates the impact of features on the output feature and identify related terms, includes the most important features and terms in the model, eliminates dubbing of the terms and features, uses different types of signal encoding, and also arranges features providing their grouping, forms a rules for precedents of previously non-experienced classes or significantly different from the existing rules of their class, adjust the existing rules based on incoming precedents, given the number previously considered observations. The proposed method allows to significantly accelerate the synthesis of neuro-fuzzy models, providing acceptable accuracy and a higher level of generalization of data, reduce complexity and redundancy, as well as increase of interpretability of neural model. The experiments on solution of practical problems of diagnosis and automatic classification are conducted. They confirmed the efficiency and applicability of the proposed method. The dependence of error of the model synthesized by the proposed method from the specified feature bitness is obtained. Using the obtained dependence allows to more rational choose the value of the number of divisions of the range of feature values in practice providing a reasonably accurate neural model. References Биргер И. А. Техническая диагностика / И. А. Биргер. – М. : Машиностроение, 1978. – 240 с. 2. Ding S. X. Model-based fault diagnosis techniques: design schemes, algorithms, and tools / S. X. Ding. – Berlin: Springer, 2008. – 473 p. 3. Интеллектуальные информационные технологии проектирования автоматизированных систем диагностирования и распознавания образов : монография / [С. А. Субботин, Ан. А. Олейник, Е. А. Гофман, С. А. Зайцев, Ал. А. Олейник] ; под ред. С. А. Субботина. – Харьков : Компания СМИТ, 2012. – 318 с. 4. Субботін С.О. Подання й обробка знань у системах штучного інтелекту та підтримки прийняття рішень : навч. посібник / С. О. Субботін. – Запоріжжя : ЗНТУ, 2008. – 341 с. 5. Computational intelligence: a methodological introduction / [R. Kruse, C. Borgelt, F.Klawonn et. al.]. – London: Springer-Verlag, 2013. – 488 p. DOI: 10.1007/978-1-4471-5013-8_1 6. Bishop C. M. Pattern recognition and machine learning / C. M. Bishop. – New York : Springer, 2011. – 740 p. 7. Buckleya J. J. Fuzzy neural networks: a survey / J. J. Buckleya, Y. Hayashi // Fuzzy sets and systems. – 1994. –Vol. 66, Issue 1. – P. 1–13. 8. Mamdani E. H. An experiment in linguistic synthesis with fuzzy logic controller / E. H. Mamdani, S. Assilian // International journal of man-machine studies. – 1975. – Vol. 7, № 1. – P. 1–13. 9. Chai Y. Mamdani Model based Adaptive Neural Fuzzy Inference System and its Application / Y. Chai, L. Jia, Z. Zhang // International Journal of Computer, Electrical, Automation, Control and Information Engineering Vol:3, No:3, 2009. –P. 663–670. 10. Леоненков А. В. Нечеткое моделирование в среде MATLAB и fuzzyTECH / А. В. Леоненков. – СПб.: БХВ-Петербург, 2003. – 736 c. 11. Химмельблау Д. Прикладное нелинейное программирование / Д. Химмельблау. – М.: Мир, 1974. – 534 с.

Anahtar Kelimeler: