Korotka L. Improving the efficiency of calculating methods of modeling the behavior of corroding constructions.

The thesis is devoted to the creation of new models of metallic constructions behavior with changing geometric characteristics as a result of corrosion and effective digital algorithms of their computer implementation with fuzzy information about environmental parameters. Object of research: modeling of the corrosion process in the loaded metal structures. The purpose of this paper is to create new behaviors corroding structures and efficient algorithms for their computer implementation for fuzzy information about the parameters of the external seredovischa. A new mathematical task statement of durability prediction and weight optimiza-tion of corroding constructions with fuzzy parameters of aggressive environment are formulated. Rate of corrosion is considered as interval value given by linguistic variable value. Aggressive environment parameter is presented as expansions in alfa-level sets. Calculating limitations of optimization task presumes numeric solution of differen-tial equations system describing the process of accumulation of geometric damages in construction elements. For the first time the method of control of its solution accuracy using artificial neural networks for determing the rational parameters of numeric integration with given maximum allowable value of error of the result is suggested. The use of neural networks significantly increases the efficiency of the algorithm by minimizing the number of integrations with solving differential equations system. A new algorithm of solving task of nonlinear mathematical programming based on joint using the adapted method of sliding tolerance and the method of deformed polyhedron is suggested for determining the optimal construction parameters. The accuracy of calculating limitations functions is adapted as a criterion of sliding tolerance. Using some former blocks of synaptic balances of neural network is allowed to propose an efficient algorithm of recalculation criterion of sliding tolerance.