Thesis

Application of neural network technologies of introscopy for determining the condition of building construction

Гафіяк, А.М.

The possibilities of introscopy technology as applied to the construction industry, for the tasks of analyzing the state of the internal structure and properties of construction objects (supports of bridges, overpasses, building foundations, road pavement, waterproofing and heat saving properties of coatings, etc.) are considered. The ultimate goal of introscopy is to establish the affiliation of the studied objects to certain groups, types, types, classes, to identify characteristic structural defects, signs of aging, loss of working properties, hidden defects, inconsistencies with technical conditions and state standards in controlled objects. The presented development of technology in the form of an ensemble of neural network models of classifiers. Classifier models have been trained, and the network learning algorithm for classifier modeling is not critical. The proposed technology may be useful in upgrading the introspection complexes with a certain refinement of the source database.

en

травень 2019 р.

building construction

Полтавський національний технічний університет імені Юрія Кондратюка

Alyoshin S.P. Application of neural network technologies of introscopy for determining the condition of building construction / S.P. Alyoshin, E.A. Borodina, A.M. Hafiiak, А.В. Nosach // Building Innovations – 2019 : зб. наук. пр. за матеріалами ІІ Міжнар. укр.-азерб. конф., 23 – 24 трав. 2019 р. – Полтава : ПолтНТУ, 2019. – С. 13-15.