MATHEMATICAL AND SOFTWARE FOR FUZZY-PRODUCTIONS KNOWLEDGE BASES GENERATION OF THE EXPERT SYSTEM IN INTEGRATED MANAGEMENT SYSTEMS

Abstract

This paper solves the problem of automating the knowledge bases generation of expert diagnostic systems. Describes the developed mathematical and software. The techniques of grouping object parameter diagnostics and constructing of a set of fuzzy production rules are offered. Constructed systems of rules is a parametric fuzzy production model of an object state. The parameters of the model are membership functions, weights conditions, and certainty factors of each rule. To identify the parameters of the model are specially designed fuzzy neural networks. In the process of learning is performed a parametric adaptation of the model to the data. The result of learning a fuzzy neural network is generated by a knowledge base of the expert system assets of fuzzy production rules with the identified parameter values. Developed software based on methods, models, and algorithms allows automating all phases of the data mining. To evaluate the efficiency of the developed mathematical software conducted researches on the approximation of the dependencies in the known data sets. The experiments showed that the resulting sets of rules for classifying the ability to have a high. Thus developed mathematical and software can be used effectively to generate knowledge bases of expert systems.