Korablyov M. Hybrid methods and models for fuzzy information processing based on artificial immune systems

The aim of the thesis is a development of theoretical foundations, development, research and improvement of methods and models that form the basis of hybrid technologies of processing fuzzy information in intelligent systems based on artificial immune systems (AIS) and increase their efficiency in the quality of decisions and the terms of results. The object of research are the processes of analyzing and processing fuzzy information The subject of research are hybrid methods and processing models of fuzzy information based on the AIS in the face of uncertainty The thesis is dedicated to solving the scientific and practical problem of developing hybrid models and methods for fuzzy information processing based on AIS, which allow more effective information analyzing by the quality of their decisions, terms of receipt and expanding a class of problems solved. The method of obtaining fuzzy expert knowledge based on the targeted procedure of incomplete pair wise comparisons is worked out. The method of determining the vector of features' priorities, coordination and adjustment of expert assessments on the basis of AIS is proposed. The methods of formalizing fuzzy expert information obtained through evaluation of quality attributes and through description of the indications of quantitative attributes in linguistic terms are highlighted. The method of objects' classification - with or without classes' standards - based on a generalized estimation of values of AF to a fuzzy set of acceptable solutions, is elaborated. An immune approach to the classification of objects in a fuzzy environment is proposed, which is characterized with using AF affinity to determine the affiliation of objects to classes. The methods of structural and parametric adaptation of fuzzy models and fuzzy neural networks based on AIS are introduced. The methods of cloning and antibodies mutation are improved. A synthesis of fuzzy controllers for coping with nonlinear dynamic objects is proposed, which involves the construction of its model, obtaining the optimal control law and adaptation of the structure and the parameters using AIS. Keywords: fuzzy model, artificial immune systems, affiliation function, adaptation, fuzzy neural network, fuzzy controller, multi-antibody, classification, identification, control.