Font Size: a A A

Study On Model Updating And Validation Technology Of Complex Mechanical Structure Based On Hierarchical Method

Posted on:2011-04-07Degree:DoctorType:Dissertation
Country:ChinaCandidate:Y ZhuFull Text:PDF
GTID:1112330362458272Subject:Mechanical design and theory
Abstract/Summary:PDF Full Text Request
Sensitivity methods based on traditional model updating technology, combined with model vali-dation in hierarchical thinking, hierarchical model proposed correction technique for a number of typical bolted frame structure, study overall technology roadmap of model validation,research model validation key -the uncertainty of the finite element model quantitative and propagation of several key issues, and to provide reliable modeling method for model prediction accuracy of the finite element response in the field of structural dynamics.The main contents of this paper are as follows:(1) Joints for more complex engineering structure, the joint parameters change greatly, the up-dating proceure is difficult when the goal of convergence, hierarchical model proposed amendments to technical amendments to the complex engineering structure, the first finite element model of the current simulation method bolts on DMIG and CHEXA unit A comparative study, showing that fully reflects the structural features of the more complex mechanical bolted frame structure, with the same level model validation of generalized mathematical model, the establishment of a hierarchical model modified based on bolt joint of the hierarchical model framework correction methods and ideas.(2) Under the proposed model in an uncertain system recognized the key issues - uncertainty modeling problem, first of all never deterministic classification, classification of the uncertain pa-rameters begin to establish the uncertainty parameters and model output of the generalized mathe-matical model; If we do not consider the model of the form of error, the error parameters for the un-certainty of modeling - to study the response surface modeling, statistical learning theory proposed by the introduction of Support Vector Machines in the response surface polynomial model to avoid the common form of abuse is difficult to choose, by A numerical example and start Garteur support vector regression models in the uncertainty modeling studies; based on hierarchical thinking of the uncer-tainty of complex engineering structures established a mathematical modeling framework, the estab-lishment of sub-structures, component parts level response surface; in particular, the structure of the error pairs a wider range of connection parameters, the fuzzy classification with Support Vector Ma-chines response surface, the principles put forward the sample simple, with fewer sample points can create accurate response surface model.(3) Error parameter uncertainty is the choice of methods to quantify key issues, proposed the use of support vector machine method of density estimation and error parameters can be approximated probability distribution, based on experimental data, with the mean square error recognition principle to establish parameters of support vector machines optimization model, crossover selection method used to obtain accurate response surface, each layer can be estimated approximate distribution of es- timates of uncertain parameters, the model for the bolted frame structure study.(4) Finite element model validation for model uncertainty in the quantification and the transmis-sion of the key issues - finite element model uncertainty comparison forward transfer and confirma-tion summarized in four aspects of the problem of research content, the experimental value of the main limits of confidence for the model validation criteria, put forward in support vector machine prediction error response to changes in working conditions and a direct mapping, combined with Monte Carlo simulation framework for the bolt connection structure of the recognized sub-domain response prediction error; point in the confirmation field - The specific operating conditions, of sub-structure uncertainty to the overall structure of the transmission connection method, draws on multidisciplinary optimization design optimization method, the overall structure without doing ex-periments, based on the sensitivity method can be passed to handle the uncertainty of the overall structure Structure.Finally, this article summarizes the contents of each chapter, highlighting the innovation and contribution to this place, proposes three-level model validation technology roadmap and future re-search work.
Keywords/Search Tags:finite element model, model updating, model validation, hierarchical model, sensitivity analysis, support vector machines, nonparametric estimation, experimental confirmation, confidence assessment, modal analysis
PDF Full Text Request
Related items