КОМПЛІМЕНТАРНІ НЕЙРОННІ МЕРЕЖІ В УПРАВЛІННІ ПОРТФЕЛЯМИ ПРОЕКТІВ ЗМЕНШЕННЯ ВТРАТ В ЕЛЕКТРИЧНИХ МЕРЕЖАХ

The model of complementary neural networks for managing project portfolios of reducing electricity consumption in electricity networks under uncertainty (observational, unverified information and imperfect methods of estimation of technological losses) is considered. The principles and methods of forming a portfolio of loss reduction projects in electric grids are defined. The properties of neural networks for parallel processing of information, self-organization, training, generalization, etc. are determined. The general model of the neural network, its elements and training techniques are given. The concepts of complementary neural network and complementarity classes are introduced. The concept of complementary neural network architecture is defined. Within the complementary networks, an algebra with a medium is introduced, which is determined by the elements of the artificial neural network and a signature that forms the classes of operations of the complementarity of the neural networks in relation to portfolio management of problems of reducing losses in electrical networks.