1 documents found
Information × Registration Number 0219U001417, 0116U001522 , R & D reports Title Theoretical grounds to estimate vulnerability of externaly perturbed thin-walled structures with the employment of non-linear deformation models and neural networks popup.stage_title Head Guk Natalya Anatoliivna, Registration Date 29-01-2019 Organization Oles Honchar Dnipro National University popup.description2 The subject of investigation is an observable thin-walled system under action of unknown emergency perturbations. The primary goal of the work is to create the computing system which identifies post-critical condition of thin-walled system and its life time at the progressive destruction caused by emergency influences. It is accomplished by preliminary identification of influence parameters by the method of inverse problems realized with the help of dynamic inverse neural networks on the basis of system supervision (measurement of deformations) with the subsequent forecast of its vulnerability on the basis of dynamic elasto-plastic calculations of the system subjected to identified influences modeled by functions of time. Research methods are analytical and computing ones. The basic scientific results are as follows. The problem of identification is formulated as correct inverse problem. The choice is carried out of the stress-strain state characteristics which are subject to measurements, and their values form compact sets. The parametrical description of constructional parameters, damages and dynamic loadings is executed. The architecture of an inverse neural network is developed and the measured parameters are chosen. The mathematical model and the computing system are constructed. Determination of time functions for the model of identification of perturbations is executed. The model of dynamic behavior of thin-walled system is formulated using elasto-plastic properties, based on the finite element method and identified parameters of the current condition which occurrence in the model of initial state is described by time functions. The module of calculations of the stress-strain state of system in a real time is created. The model of calculation of vulnerability is constructed and its program realization is carried out. The forecasting model of life time is created on the basis of reference chaotic time series. The module realizing chaotic time series on the basis of the information about post critical forms of deformation is developed. Program realization of the forecast of vulnerability is executed. The forecast assumptions concerning development of received results can be used during monitoring of real bearing ability and life time of thin-walled systems during service. Practical value of work will consist in creation of theoretical bases for estimation of vulnerability which can be used for making decisions about further exploitation of damaged systems. The received results can be used during the analysis of operational opportunities of aviation designs, and also in system of antimissile defense. A problem of residual operability is of great importance in the engineering industry and building industry. The damage estimation according to the mentioned fracture criterion leads to resource-saving in these industries due to enlargement of the time periods between general overhauls. Proposed techniques could be used in space (Yuzhnoe Design Office), aviation (Progress Design Office Motor-Sich Ltd.), and engineering (Malyshev Plant Ltd.) industries. Product Description popup.authors Адлуцький Віктор Якович Громов Василь Олександрович Козакова Наталія Леонідівна Магас Олексій Сергійович Ободан Наталія Іллівна Полішко Олексій Миколайович Радовський Михайло Петрович Сіліч-Балгабаєва Валентина Борисівна Сохач Юрій Васильович Фрідман Олександр Давидович Шаповал Ірина Павлівна Шевельова Алла Євгенівна popup.nrat_date 2020-04-02 Close
R & D report
Head: Guk Natalya Anatoliivna. Theoretical grounds to estimate vulnerability of externaly perturbed thin-walled structures with the employment of non-linear deformation models and neural networks. (popup.stage: ). Oles Honchar Dnipro National University. № 0219U001417
1 documents found

Updated: 2026-03-21