Kucherenko D. Models and quality improvement methods of computer systems' expert diagnosis using production rules

In the thesis the models and analysis methods of the logical correctness of production rules in fuzzy inference systems by checking them for completeness, consistency, coherence, and minimality were proposed, that made it possible to formalize the processes of knowledge base formation and analysis for improving the quality of computer systems expert diagnosis; the representation model of expert knowledge in diagnosis expert systems as a production rules in a cubic form, which is characterized by qubit encoding of multi-valued cubic alphabet characters,was improved, that made it possible to use logical operations, and to accelerate computation greatly during production rules analysis; the model of diagnostic features representation of various components of computer systems in a form of linguistic variables through the establishment of a formalized template of their representation and estimation was further developed, which allows using the unified approach to construction of an expert knowledge base and expanding a list of diagnostic features in diagnosis expert systems; the models and synthesis methods of fuzzy inference production rules in a cubic form by assigning weights for each term of input linguistic variable were further developed, which allow an expert to formalize the process and reduce the time for creating production rules with an arbitrary number of diagnostic features and ranges of their values.