Yakymets N. Methods and tools for development of fault tolerant digital control systems using genetic algorithms

Subject of research - development process and reliability of digital control systems for airspace and other critical applications. Goal of research - increasing fault tolerance of digital control systems with programmable logic (DCSPL) by developing and implementing methods and tools for synthesising such systems with genetic algorithms. Pursued researches are based on methods of automatic control theory, digital automata theory, theory of algorithms and genetic algorithms while developing models of fault tolerant DCSPL with partially definite and partially correct automata; methods of probability theory and reliability theory while developing evaluation reliability models of DCSPL based on partially definite and partially correct automata. In the thesis, models, methods and tools for synthesis of DCSPL with partially definite and partially correct automata are proposed. The scientific results include developing the models of fault tolerant digital control systems with programmable logic based on partially functioning automata, which difference from the known ones is that the structure of the latter models is controlled with information about the definiteness and correctness of each automaton in mind; the method of developing multi-version fault tolerant digital control systems with programmable logic improved in a part of the receiving versions flow using genetic algorithms that allow to automate the synthesis of different system versions with the required features; the normalized rank selection method that allow reducing time constrains for developing versions of digital control systems with programmable logic by normalizing the probabilities of individual selection. Degree of application - industrial. Main results can be used in organizations and enterprises engaged in development and implementation of specialized digital devices.