Orientador: Paulo Roberto Gardel Kurka / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Mecanica / Made available in DSpace on 2018-07-24T15:32:08Z (GMT). No. of bitstreams: 1 Mathias_MauroHugo_D.pdf: 7532948 bytes, checksum: 3f7d919ebab4f2d9d68fbf29915d517a (MD5) Previous issue date: 1998 / Resumo: Esta tese trata do estudo de teorias, estratégias e metodologias de identificação de sistemas aplicado à análise de sinais de estruturas mecânicas. A abordagem do problema de identificação no contexto das áreas afins de análise de sinais, análise de séries temporais e análise modal é também discutida no trabalho. O enfoque central da pesquisa consiste em aplicar técnicas paramétricas do domínio do tempo na análise modal de estruturas mecânicas. Os procedimentos de identificação adotados descrevem o sistema estrutural através de modelos derivados do método Auto Regressivo de Médias Móveis (ARMA), correlacionando os parâmetros modais e coeficientes do modelo. Testes de simulação foram desenvolvidos para demonstrar as características dos métodos de identificação estudados. Finalmente um modelo modal experimental foi desenvolvido. Os parâmetros modais identificados do modelo foram estabelecidos a partir dos sinais de excitação e resposta medidos na estrutura através do procedimento experimental / Abstract: This thesis is related to system identification theories, strategies and methodologies applied to signal analysis of mechanical structures. The approach of the identification problem in the setting of signal analysis, time series models and modal analysis was also discussed in the work. The central subject of the research is outlined on the use of parametric time domain techniques applied to mechanical structural modal analysis. The identification procedures describes the structural system by models derived from an Auto Regressive Moving Average method (ARMA), highlighting the relations between modal parameters and model coefficients. Simulated test cases have been developed to demonstrate the characteristics of the identification methods studied. Finally an experimental modal model has been developed. The identified modal parameters were established from the excitation and response signals measured on the structures through the experimental test / Doutorado / Mecanica dos Sólidos e Projeto Mecanico / Doutor em Engenharia Mecânica

Orientador: Paulo Roberto Gardel Kurka / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Mecanica / Made available in DSpace on 2018-08-04T14:31:42Z (GMT). No. of bitstreams: 1 AmaroBaldeon_RobertoPedro_D.pdf: 4773363 bytes, checksum: d0f431ee3b99fa29f96a302fb22a01c1 (MD5) Previous issue date: 2004 / Resumo: O trabalho apresenta os modelos de series temporais aplicados na identificação de sistemas mecânicos. A técnica de Máxima Verossimilhança (ML) é aplicada na estimação dos parâmetros dos modelos de series temporais a fim de melhorar a precisão na estimação dos parâmetros modais. O algoritmo de Spliid é usado na estimação de valores iniciais para iniciar o processo iterativo da técnica ML. A performance da técnica ML é verificada simulando um sistema de três graus de h"berdade com duas entradas e duas saídas, é adicionado ruído no sinal de saída a fim de se verificar a eficiência dos modelos de series temporais na presença de ruído estocástico. Dados provenientes de teste de vibração de um protótipo de asa de avião são usados para comparar as técnicas propostas com os resultados obtidos usando o Ideas/Test / Abstract: This work presents the Multivariate Time Series Models in the identification ( mechanica1 systems. The maximum likelihood technique (ML) is applied to estimate tl1 parameters of the time series models in order to improve the precision in the estimation ( modal parameters. The Spliid's algorithm is used to estimate initial values to start tt iterative process of the ML technique. The perfoimance of the ML technique is verified j a three degree fteedom simulated system with two inputs and two outputs. Stochastic noh is added to the outputs in order to verify the performance of time series models when ti output is influenced by stochastic noise. Vibrating test data of a wing prototype is used 1 compare the proposed techniques with the results obtained experimentaly from Ideas/ Test / Doutorado / Mecanica dos Sólidos e Projeto Mecanico / Doutor em Engenharia Mecânica

[pt] A estabilidade de tensão tornou-se nos últimos anos tema de intensa pesquisa, envolvendo principalmente a identificação da natureza do problema e a busca de estratégias mais favoráveis para a operação e planejamento dos sistemas de potência. Este trabalho apresenta um relato da evolução da análise do fenômeno sendo, posteriormente descritos os detalhes de sua abordagem através da teoria das bifurcações em sistemas dinâmicos não-lineares. São caracterizados o limite de transmissão de potência à estabilidade de tensão e a geometria das trajetórias no espaço de estados em pontos de equilíbrio em suas proximidades. O estabelecimento de equivalência entre os modelos dinâmico e estático tornou possível o emprego de métodos baseados nas equações algébricas do fluxo de carga na determinação do limite de estabilidade de tensão e na avaliação do comportamento dinâmico local. Com base no algoritmo de fluxo de carga convencional, foi desenvolvido um programa computacional que emprega o método de continuação parametrizado. Os métodos de continuação permitem a obtenção do conjunto completo das soluções reais, relacionadas por um parâmetro escalar, de um sistema de equações. Mostra-se que a aplicação desses métodos em conjunto com as técnicas de análise modal resulta em uma forma eficiente para a seleção e localização de reforços visando aumentar a capacidade global de transmissão dos sistemas de potência. / [en] In the past few years voltage stabiity has become a subject of intense research involving the identification of the nature of this problem and the seek of suitable strategies for power systems operation and planning. In this work it is presented a report on the evolution in the analysis of this phenomena and it is described the details of its approach through the dynamic nonlinear systems bifurcation theory. The power transmission limit imposed by the voltage stability conditions is addressed and the geometry of the trajectories in state space considering the equilibrium points on the vicinity of this limit is deeply studied. The establishment of the equivalence conditions between the dynamic and static power system models made it possible to employ load-flow algebraic equations based methods on both the determination of the power transmission limit and the evaluation of local dynamic behaviour. A load-flow based computer program which uses the parameterized continuation method was developed. The continuation methods are widely known for providing the determination of whole set of a system s solutions which are related by a scalar parameter. In the closure, it is shown that the application of these methods, in a complementary basis with the modal analysis techniques, results in an efficient procedure to promote the choice and location of devices to improve the power system global tranmission capacity.

In this thesis a methodology for measuring vibrations has been produced andinvestigated for APID 60, a rotorcraft in a Vertical Take-off and landing remotelypiloted aircraft system (VTOL RPAS). A comparative study was carried out forthe purpose of identifying the methodology with respect to design modificationscommon to the APID 60. The pilot-study identified experimental modal analysis(EMA) as a feasible part of the methodology for experimentally extracting themodal parameters of a structure. The EMA was performed on the main frameof the APID 60 where an impact hammer test was chosen as the technique forextracting the response data. As a comparison a point mass was added to thestructure to alter the dynamic properties and the test was repeated.The results from the EMA was compared with a modal analysis performednumerically with a calculation software. Comparison of the results from EMAwith the modal analysis performed numerically indicates consistency. This confirmsa good reliability of the methodology produced. However, the structure onwhich the test were preformed is simple in terms of constant structural properties.Further work should therefore investigate whether this methodology of measuringvibrations could be successfully applied to a structure with higher complexity.

All structures have modal properties such as natural frequencies and damping. In engineeringit is often of interest to estimate these modal properties for certain structures, to be used whenmodelling for example fatigue. This is done by computing them from finite element models, by using experimental measurements or both. In the case of doing both, a finite elementmodel is usually established first and adjusted to fit measurements from experiments. Atlas Copco Rock Drills AB is the company where this thesis has been performed and the subject is experimental procedures related to estimating modal properties of the so calledpower pack, which essentially is a modularised engine and hydraulic power source of an Atlas Copco drill rig. Their current method for estimating these properties is a classical procedure which makes use of an impact hammer that an operator strikes the power pack with to induce excitation. Due to concealment of behind other parts the power pack when mounted inside the drill rig, the number of places where the operator is able to strike the power pack in is limited. Another problem with the current procedure is that it can be difficult to strike the power pack with a strong enough impulse to generate reliable results. In this thesis a new experimental procedure for Atlas Copco to use is suggested. It is based on operational modal analysis (OMA), which uses the machinery's excitation from its operational conditions to compute modal estimates. A comparison between different experimental procedures have been done and the suggested procedure is the following: excitation by engine sweep; modal identifcation by the PolyMAX method and mode shape scaling by the harmonic scaling method. An experiment was performed to compare two OMA procedures.The suggested procedure is the one that generated the better results of the two.

La détermination des caractéristiques vibratoires est une étape cruciale lors du processus de certification d'une structure ou lors de l'évaluation de sa détérioration. Quelle que soit la technique utilisée, l'identification des modes d'une structure nécessite d'effectuer des essais de vibration. Durant ces essais, on effectue des mesures qui permettent d'observer la réponse du système à une excitation connue, partiellement connue ou inconnue. Les méthodes d'identification modale sont des méthodes globales. Elles permettent de déterminer les modes de la structure à partir d'un post-traitement des données mesurées. Certaines techniques d'identification modale, telles que les méthodes basées sur la minimisation d'erreur de modélisation ARX ou les méthodes sous-espaces, se sont fortement répandues dans le monde industriel ces quinze dernières années. Cependant, malgré ce succès, ces méthodes ne fourniront jamais les paramètres modaux exacts de la structure étudiée. En effet, à cause de nombreuses sources d'erreurs (e.g. bruits sur les capteurs, bruits environnementaux), l'identification modale à partir de données réelles ne peut mener qu'à des estimations des paramètres modaux. Une estimation est par définition entachée d'une erreur statistique, la variance. Le but de cette thèse est l'étude approfondie de l'estimation de cette erreur statistique. Le manuscrit est découpé en trois partie. La première concerne la description détaillée de deux familles d'identification modale (minimisation d'erreur de modélisation ARX et méthodes sous-espace). La seconde partie décrit le principe de l'estimation des incertitudes et fournit les chaînes de propagation des variances. La troisième partie présente une validation poussée de ces méthodes dans le but d'une utilisation opérationnelle. / Determinination of vibration characteristics is a crucial step for structure certification or health monitoring. Whatever the method, modes identification of a structure requires vibration tests. During these tests, measurements are made in order to observe responses of the system to known, partially known or unknown excitations. Modal identification methods are global methods. They allow the determination of modes from data post-treatment. Some modal identification methods, such as those based on ARX modelisation error minimisation or subspace methods, have been largely spread into the industry for about fifteen years. Despite this succes, these methods will never provide exact results. Indeed, due to numerous error sources (e.g. sensor noises, environmental noises), modal identifications based on measured data can only lead to modal parameters estimations. By definition, an estimation is subject to statistical errors, also called variances. The purpose of this thesis is to give an in-depth study of these statistical errors estimation. This document is divided in three parts. Firstly, a description of two kind of modal identification methods (ARX modelisation error minimisation and subspace methods) is given. Secondly, the uncertainty estimation principle is explained and propagation schemes of variances are given. Finally, the third part provides an in-depth validation of these methods in order to proceed to industrial utilisations.

In this work, a pioneering application of the Solution Structure Method (SSM) for structural dynamics problems is presented. Vibration analysis is an important aspect of any design-analysis cycle for which reliable computational methods are required. Unlike many meshfree methods, SSM is capable of {\it exact treatment of all prescribed boundary conditions}. In addition, the method is capable of using basis functions which do not conform to the shape of the geometric model. Together, this defines an unprecedented geometric flexibility of the SSM. This work focused on the development of numerical algorithms for 2D in-plane and 3D natural vibration analysis and 2D in-plane dynamic response. The convergence and numerical properties of the method were evaluated by comparing meshfree results with those obtained using traditional Finite Element Analysis implemented in Solidworks and ANSYS. The numerical experiments presented in this work illustrate that the Solution Structure Method possesses good convergence and in some cases, such as geometries with partially fixed boundaries, this method converges much more rapidly than traditional FEA. Finally, in addition to complex boundary conditions, this method can easily handle complex geometries without losing favorable convergence properties.

This master´s thesis deals with the acoustic emission of the gearbox housing. It contains a description of the procedure of solving individual numerical approaches and their results. Finally, it compares the influence of design modifications on acoustic emission.

No description available.

The master thesis deals with computational modeling of a turbocharger vibrations and and assessment of influnce of passive dynamic vibration absorber on vibrations of actuator bracket. The use of dynamic vibration absorber was summarized in the research study. The analysis were performed using finite element method in ANSYS. Several computational models of turbocharger were created with different ways of modeling bolted joints between turbocharger parts. Modal analysis of each model was performed and the results were compared. For the selected model, the response to the kinematic excitation from the internal combustion engine for two load conditions was calculated using harmonic analysis. A simple model of vibration dynamic absorber was applied to the turbocharger model with reduced degrees of freedom and its influnce on vibrations of actuator bracket was investigated. Significant decrease of the maximum acceleration amplitude was achieved in a given frequency range when absorber parameters were optimized.