Industrial Automatic Control Systems and Controllers Annotation << Back Research of the Possibility of Prediction of the Resource of Composite Structures Dased on the Results of Optical-thermal Nondestrousing Control Using Deep Neural Networks S.O. Kozelskaya The paper considers the problem of increasing the reliability of assessing the maximum service life of multicomponent structures by constructing predictive models using the results of optical-thermal non-destructive testing of the state of objects by temperature fields and the value of internal deformation of the material under force action on the structure as input information. It is shown that in the case of using logical approaches as a software tool for predicting the ultimate resource of structures made of polymer composite materials, part of the knowledge should be used for reasoning that provides an explanation of the conclusions drawn, since formal logic is of limited applicability, especially in conditions of incomplete or uncertain information. In this case, the solution to the problem becomes the identification and establishment of cause-andeffect relationships. For the tasks of technical assessment of the quality of structures and their service life, the use of such logical conclusions as inductive, deductive and analogous conclusions is impossible, since for their work, they require all information about the diagnosed structure. Keywords: safety of structures; forecasting, artificial intelligence; forecasting the resource of structures; deep neural network; polymer composite materials; flaw detection; incomplete information; non-destructive testing; data processing. DOI: 10.25791/asu.1.2021.1252 Pp. 44-56.