CASPIAN JOURNAL
MANAGEMENT AND HIGH TECHNOLOGIES
NEURO-FUZZY MODEL AND SOFTWARE COMPLEX FOR FORMING KNOWLEDGE BASES FOR OBJECTS STATE ASSESSING
 |
Read | Kataseva Dina V., Ismagilov Ilyas Idrisovich NEURO-FUZZY MODEL AND SOFTWARE COMPLEX FOR FORMING KNOWLEDGE BASES FOR OBJECTS STATE ASSESSING // Caspian journal : management and high technologies. — 2022. — №1. — pp. 65-76. |
Kataseva Dina V. - postgraduate student, Kazan National Research Technical University named after A.N. Tupolev-KAI, dvkataseva@kai.ru
Ismagilov Ilyas Idrisovich - Kazan National Research Technical University named after A.N. Tupolev-KAI
The task of objects state assessing is described. To improve the efficiency of its solution and reduce the human factor influence, the expediency of using decision-making support systems and the formation of fuzzy-production type knowledge bases has been updated. As a tool for their formation, it is proposed to use a neuro-fuzzy model based on a fuzzy neural network training. The features of the initial data for training are considered. A type of fuzzy-production rules that make up the knowledge base is proposed. The stages of neuro-fuzzy model construction, including the initialization and setting of the values of fuzzy neural network parameters, are considered. The developed method of its training based on the use of a genetic algorithm is described. For its implementation, the issues of choosing and coding learning parameters are considered on the example of triangular membership functions. A fitness function and a criterion for choosing the best chromosome in the genetic algorithm are proposed. Based on the proposed methods, a software package for the knowledge bases formation has been developed. Its structure, composition of program modules and their functionality are presented. The results of the software package approbation are presented on the example of the knowledge base formation for the selection of geological and technical measures in oil fields. The approbation results confirmed the effectiveness of the proposed approach and the possibility of its use in the knowledge bases formation and the decision-making support systems construction for objects state assessing.
Key words: object state assessment, knowledge base, fuzzy neural network, neuro-fuzzy model, fuzzy production rule, membership function, genetic algorithm, decision support system, geological and technical measures, oil field