Moroz O. Combinatorial-genetic method for inductive construction of models of complex objects

The thesis is devoted to developing a method for constructing models based on hybridization between combinatorial GMDH algorithm and genetic algorithm for solving inductive modeling problems. For the first time, the self-organizing genetic search of the optimal model structure is applied in a sorting-out GMDH algorithm allowing to solve large-dimensional modeling problems. The operation speed, accuracy and stability of the hybrid algorithm are investigated. A computer technology was developed and implemented for constructing models of complex objects from observation data. The application results of the technology to solve three problems are presented: modeling abilities of players in real-time strategy (RTS) games, modelling of changes in prices for apartments in buildings being designed, modeling of seasonal changes of amylolytic microorganisms quantity on copper polluted plot of soil and reconstruction of missing observation data.