301 |
A Study of Quadrotor ModellingSonntag, Dag January 2011 (has links)
Quadrotors are a type of aircraft that lately has gained increased popularity within the UAV-scientific area. Many research groups around the world have implemented control systems that allow for autonomous flight of quadrotors with the help of the known dynamics. This thesis presents two approaches to modelling the dynamics of the quadrotor. The first is a linear greybox approach where the structure is derived from known equations and some constants are measured and some identified through system identification techniques. The second model is a blackbox model where a neural network is trained and used. The two models are then evaluated using known error measurements with the help of previously recorded flight data and the results are presented. It is for example shown that with the untreated flight data the traditional greybox model have accurate dynamics but is sensitive to noise and drifts in the measurements. It is also shown that better results generally can be achieved using a neural network model, especially for noisy unpreprocessed data.
|
302 |
Modeling and Control of Lime Addition in a Flotation ProcessTammia, Rasmus January 2017 (has links)
Flotation is an important and versatile mineral processing technique that is used to separate hydrophobic materials from hydrophilic. This technique makes it possible to mine complex ores that otherwise would have been regarded as uneconomic and non-beneficial. In this case flotation is used to separate copper from the unwanted gangue. The addition of lime is used to control the pH level in the flotation’s pulp, which governs the selectivity of the process, i.e. which minerals are recovered. Currently, fluctuating concentration grades of the produced metals have been observed in Boliden Aitik. Therefore, Boliden proposes a new control strategy which aims to maintain a constant ratio between the added lime and the incoming ore flow, but at the same time ensuring that the pH level is maintained within allowed limits. The aim of this thesis is to develop a model that captures the most essential dynamics of a process stage where lime is added, and then evaluate the suggested control strategy by studying suitable control structures. A linear model describing the system dynamics in a specific operating region is obtained by conducting step response experiments on the process. The model is then used to obtain a model describing the disturbances of the process, thereby yielding a complete model that describes the most important dynamics. The most promising control structure utilizes the concept of selective control, where a ratio controller is allowed to maintain a constant ratio as long as the pH level is within allowed boundaries. The pH level is maintained within the boundaries with upper and lower bound pH controllers that utilize the concept of an equivalent control objective (known as the strong acid equivalent) in order to achieve satisfying pH control. The results show that the control structure is able to maintain a constant ratio, and also ensure that the pH level is kept within the allowed limits. A cascade inspired pH ratio controller is also studied and evaluated. The results show that this pH ratio controller is only able to maintain a constant ratio as long as the incoming ore flow is constant. However, the outcomes also suggest that the concentration grades are either sensitive to variations in the ratio between added reagent and incoming ore flow, or that there is something else that causes them to vary.
|
303 |
The identification of unbalance in a nonlinear squeeze-film damped system using an inverse method : a computational and experimental studyTorres Cedillo, Sergio Guillermo January 2015 (has links)
Typical aero-engine assemblies have at least two nested rotors mounted within a flexible casing via squeeze-film damper (SFD) bearings. As a result, the flexible casing structures become highly sensitive to the vibration excitation arising from the High and Low pressure rotors. Lowering vibrations at the aircraft engine casing can reduce harmful effects on the aircraft engine. Inverse problem techniques provide a means toward solving the unbalance identification problem for a rotordynamic system supported by nonlinear SFD bearings, requiring prior knowledge of the structure and measurements of vibrations at the casing. This thesis presents two inverse solution techniques for the nonlinear rotordynamic inverse problem, which are focused on applications where the rotor is inaccessible under operating conditions, e.g. high pressure rotors. Numerical and experimental validations under hitherto unconsidered conditions have been conducted to test the robustness of each technique. The main contributions of this thesis are:• The development of a non-invasive inverse procedure for unbalance identification and balancing of a nonlinear SFD rotordynamic system. This method requires at least a linear connection to ensure a well-conditioned explicit relationship between the casing vibration and the rotor unbalance via frequency response functions. The method makes no simplifying assumptions made in previous research e.g. neglect of gyroscopic effects; assumption of structural isotropy; restriction to one SFD; circular centred orbits (CCOs) of the SFD. • The identification and validation of the inverse dynamic model of the nonlinear SFD element, based on recurrent neural networks (RNNs) that are trained to reproduce the Cartesian displacements of the journal relative to the bearing housing, when presented with given input time histories of the Cartesian SFD bearing forces.• The empirical validation of an entirely novel approach towards the solution of a nonlinear inverse rotor-bearing problem, one involving an identified empirical inverse SFD bearing model. This method is suitable for applications where there is no adequate linear connection between rotor and casing. Both inverse solutions are formulated using the Receptance Harmonic Balance Method (RHBM) as the underpinning theory. The first inverse solution uses the RHBM to generate the backwards operator, where a linear connection is required to guarantee an explicit inverse solution. A least-squares solution yields the equivalent unbalance distribution in prescribed planes of the rotor, which is consequently used to balance it. This method is successfully validated on distinct rotordynamic systems, using simulated data considering different practical scenarios of error sources, such as noisy data, model uncertainty and balancing errors. Focus is then shifted to the second inverse solution, which is experimentally-based. In contrast to the explicit inverse solution, the second alternative uses the inverse SFD model as an implicit inverse solution. Details of the SFD test rig and its set up for empirical identification are presented. The empirical RNN training process for the inverse function of an SFD is presented and validated as a part of a nonlinear inverse problem. Finally, it is proved that the RNN could thus serve as reliable virtual instrumentation for use within an inverse rotor-bearing problem.
|
304 |
Open and closed loop model identification and validationGuidi, Figuroa Hernan 03 July 2009 (has links)
Closed-loop system identification and validation are important components in dynamic system modelling. In this dissertation, a comprehensive literature survey is compiled on system identification with a specific focus on closed-loop system identification and issues of identification experiment design and model validation. This is followed by simulated experiments on known linear and non-linear systems and experiments on a pilot scale distillation column. The aim of these experiments is to study several sensitivities between identification experiment variables and the consequent accuracy of identified models and discrimination capacity of validation sets given open and closed-loop conditions. The identified model structure was limited to an ARX structure and the parameter estimation method to the prediction error method. The identification and validation experiments provided the following findings regarding the effects of different feedback conditions: <ul> <li>Models obtained from open-loop experiments produced the most accurate responses when approximating the linear system. When approximating the non-linear system, models obtained from closed-loop experiments were found to produce the most accurate responses.</li> <li>Validation sets obtained from open-loop experiments were found to be most effective in discriminating between models approximating the linear system while the same may be said of validation sets obtained from closed-loop experiments for the nonlinear system.</li> </ul> These finding were mostly attributed to the condition that open-loop experiments produce more informative data than closed-loop experiments given no constraints are imposed on system outputs. In the case that system output constraints are imposed, closed-loop experiments produce the more informative data of the two. In identifying the non-linear system and the distillation column it was established that defining a clear output range, and consequently a region of dynamics to be identified, is very important when identifying linear approximations of non-linear systems. Thus, since closed-loop experiments produce more informative data given output constraints, the closed-loop experiments were more effective on the non-liner systems. Assessment into other identification experiment variables revealed the following: <ul> <li>Pseudo-random binary signals were the most persistently exciting signals as they were most consistent in producing models with accurate responses.</li> <li>Dither signals with frequency characteristics based on the system’s dominant dynamics produced models with more accurate responses.</li> <li>Setpoint changes were found to be very important in maximising the generation of informative data for closed-loop experiments</li></ul> Studying the literature surveyed and the results obtained from the identification and validation experiments it is recommended that, when identifying linear models approximating a linear system and validating such models, open-loop experiments should be used to produce data for identification and cross-validation. When identifying linear approximations of a non-linear system, defining a clear output range and region of dynamics is essential and should be coupled with closed-loop experiments to generate data for identification and cross-validation. / Dissertation (MEng)--University of Pretoria, 2009. / Chemical Engineering / unrestricted
|
305 |
Aplicação de técnicas de identificação paramétricas e não paramétricas em uma bancada de vibração torcional / Applications of parametric and nonparametric identification techniques on a torsional vibration setupPuerto Acosta, Jorge Andres, 1979- 12 April 2013 (has links)
Orientador: Juan Francisco Camino / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Mecânica / Made available in DSpace on 2018-08-24T10:56:37Z (GMT). No. of bitstreams: 1
PuertoAcosta_JorgeAndres_M.pdf: 7253811 bytes, checksum: c3adc61ec5688b2b451d7b3b0643522e (MD5)
Previous issue date: 2013 / Resumo:
Esta dissertação apresenta a aplicação de técnicas de identificação paramétricas e não paramétricas para obter o modelo linear invariante no tempo de uma bancada de vibração torcional eletromecânica. A bancada é composta por um servomotor e um disco acoplado ao eixo do motor, além de dois freios eletromagnéticos alocados no disco. A bancada representa múltiplos sistemas giratórios usados na indústria, como, por exemplo, os eixos giratórios em tornos com controle numérico computadorizados. A bancada se encontra no Laboratório de Controle e Identificação da Faculdade de Engenharia Mecânica da UNICAMP. Com essa bancada se realizam pesquisas na área de controle e identificação, sendo assim necessário ter um modelo validado experimentalmente. O processo de obtenção do modelo divide-se em três etapas: [Etapa1] Revisão dos aspectos conceituais de identificação, configuração física da bancada, análise de ruído e tratamento dos sinais e análise preliminar das faixas de operação da bancada. [Etapa 2] Aplicação dos métodos de identificação estudados anteriormente, com detalhes técnicos do processo de identificação. Na identificação da bancada de vibração torcional são propostos dois modelos para o acoplamento: (i) corpo com rigidez infinita (rígido) e (ii) corpo com uma rigidez finita (flexível). Para o modelo cujo o acoplamento é rígido, se propõe uma função de transferência de primeira ordem e para o caso do acoplamento flexível, se propõe uma função de transferência de terceira ordem. Para as duas hipóteses, modelos de primeira e terceira ordem são obtidos usando-se os métodos de identificação ARX, OE , N4SID e PEM. Os modelos identificados são comparados com a estimação da FRF obtida experimentalmente. Também é comparada a resposta temporal dos modelos identificados com os dados experimentais. Dessas identificações, um modelo inicial é escolhido. Do modelo inicial são obtidos os parâmetros iniciais a serem usados na identificação paramétrica de caixa cinza. Identificação na qual são obtidos os parâmetros dos modelos. Os modelos, são identificados para cinco condições de operação diferentes. Esses cinco modelos lineares invariantes no tempo (LTI) são denominados modelos locais. Cabe salientar que cada condição de operação, depende da variação do amortecimento do disco na bancada. O amortecimento depende da corrente aplicada nos freios eletromagnéticos. [Etapa 3] Processo de validação experimental dos modelos obtidos. A validação é feita apresentando um processo recursivo. Tendo como critério de validação o melhor ajuste de curva dado pelo menor erro. Como resultado da aplicação das etapas e processos contidos neste trabalho, se obtém um modelo validado experimentalmente para o caso em que o acoplamento possui rigidez infinita. Da mesma maneira é obtido o modelo validado experimentalmente para o caso em que o acoplamento possui uma rigidez finita / Abstract: This dissertation presents an application of parametric and nonparametric identification techniques to obtain a linear model of a electro-mechanical torsional vibration bench. The torsional vibration bench consist of a servomotor with a disk attached to the motor's shaft. The torsional vibration bench can represents multiple systems used in the industry, as the rotating shafts in machines, for example, computerized numerical control machines. The torsional vibration bench is located in the School of Mechanical Engineering, UNICAMP. This torsional vibration bench is used for research in control and identification. Thus, it is necessary to have a validated experimental model of the torsional vibration bench. The process to obtain a model is divide in three step: [Step 1] is a preliminary part that contains: identification concepts, the torsional vibration bench configuration, analysis of noise and signal¿s processing, and analysis of a rank of frequency's work and excitation's torque, for the torsional vibration bench. [Step 2] is the application of the identification methods studied in step 1, giving the technical details of the identification process. The following hypothesis are used for the coupling: (i) the coupling is a body with infinite stiffness; (ii) the coupling is a body with finite stiffness. For five operating conditions, linear time invariant models are obtained, which are called local models. It should be noted that each operating condition depends on the variation of the damping of the disk on the bench. The damping depends on the fixed current in the electromagnetic brakes. A first-order transfer function is proposed for the system with rigid coupling and a third-order transfer function is proposed for the model with flexible coupling. The models identified using ARX, OE, N4SID, and PEM techniques are compared with the estimation of the FRF. The temporal response of the models identified are compared with the experimental data. [Step 3] Is the experimental validation process used to validate the local models. As a result of applying the steps and processes contained in this work, a model is validated experimentally for the case in which the coupling has infinite stiffness. Likewise, one validated model is obtained for the system in which the coupling has finite stiffness. In the parametric identification case a set of validated models are identified experimentally using time domain methods / Mestrado / Mecanica dos Sólidos e Projeto Mecanico / Mestre em Engenharia Mecânica
|
306 |
Detecção e localização de falhas via observadores de estado e de ordem reduzida / Faults detection and isolation using reduced-order state observersMelo, Gilberto Pechoto de, 1963- 14 August 1998 (has links)
Orientador: Robson Pederiva / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Mecanica / Made available in DSpace on 2018-07-24T02:53:53Z (GMT). No. of bitstreams: 1
Melo_GilbertoPechotode_D.pdf: 10688875 bytes, checksum: fb744df0fd371e40e7782c8c85921d94 (MD5)
Previous issue date: 1998 / Resumo: Um dos fatores do grande interesse no desenvolvimento de novas técnicas de detecção de falhas é devido ao aumento da demanda da indústria em relação a segurança de seus sistemas, sendo eles supervisionados e monitorados para que as falhas sejam sanadas o mais rápido possível e que os distúrbios em operação normal não causem uma deterioração da performance dos mesmos. Neste trabalho, desenvolveu-se uma metodologia para detecção e localização de falhas em sistemas mecânicos utilizando observadores de estado de ordem reduzida. O método pode reconstruir os estados não medidos ou os valores provenientes de pontos de difícil acesso no sistema. Os parâmetros de interesse sujeitos a falhas são escolhidos, projetando-se um observador global otimizado para análise de todo o sistema considerando possíveis perturbações aleatórias na excitação, na resposta e falhas nos sensores. Projeta-se também observadores robustos a estes parâmetros de interesse, que localizam possíveis falhas ou irregularidades no sistema. Para os componentes que necessitem de um acompanhamento periódico devido às suas grandes solicitações ou falhas constantes, montam-se observadores com um sistema de alarmes que gera uma curva de tendências em um sistema automático para detecção e localização de falhas desenvolvido neste trabalho. Para os sistemas simulados e experimental, fez-se análises da performance transiente e em regime permanente, excitando-se os sistemas com força impulsiva, força senoidal, ruído aleatório, combinações dos mesmos, etc. Os resultados foram bons quando comparados com outros métodos e pôde-se também verificar os resultados através de uma bancada de testes / Abstract: The development of new faults detection techniques is necessary because of increasing demands from industries on reliability and safety in mechanical systems. They must be supervised such that occurrence of failures can be accommodated as quickly as possible because they can cause an unacceptable deterioration of the systems performance. In this work, we have developed a methodology to Detect and Isolate Faults in mechanical systems using reduced order state observers. We can monitor unmeasureble variables and the method selects the parameters from components that may fault during the process and constmcts an otimized global observer to analyze alI the system considering random noises in the excitation, in the response and sensor faults. To isolate component failures via robust observation, an automatic system with a bank of detection observers is constmcted, where each observer is only sensitive to one specified component failure while robust to alI other component failures. We have analized the transient and steady-state performance by exciting the system with impulsive force, sinosoidal force, random noise etc. The results were good when we compare them with other methods and we have verified the results through a testing rig / Doutorado / Mecanica dos Sólidos e Projeto Mecanico / Doutor em Engenharia Mecânica
|
307 |
Estimação On-Line de parâmetros dependentes do estado (State Dependent Parameter - SDP) em modelos de regressão não lineares / State dependent parameters (SDP) On-line estimation for nonlinear regression modelsAlegria, Elvis Omar Jara, 1986- 27 August 2018 (has links)
Orientador: Celso Pascoli Bottura / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação / Made available in DSpace on 2018-08-27T02:15:32Z (GMT). No. of bitstreams: 1
Alegria_ElvisOmarJara_M.pdf: 5581682 bytes, checksum: cd5b08b04c7ba4bcd505ab00e5335ffc (MD5)
Previous issue date: 2015 / Resumo: Este trabalho é sobre a identificação recursiva em tempo real das dependências parâmetro-estado em modelos de regressão de series temporais estocásticas. O descobrimento dessas dependências é útil para obter uma nova, e mais acurada, estrutura do modelo. Os métodos recursivos convencionais de estimação de parâmetros variantes no tempo, não conseguem bons resultados quando os modelos apresentam parâmetros dependentes do estado (SDP) pois eles tem comportamento altamente não linear e inclusive caótico. Nossa proposta está baseada no estudo de Peter Young para SDPs no caso Off-Line. É discutido o método que ele propõe para reduzir a entropia das séries nos modelos com SDP e para isto se apresenta umas transformações dos dados. São propostas mudanças no seu algoritmo Off-Line que o fazem mais rápido, eficiente e manejável para a implementação do modo On-Line. Finalmente, três exemplos numéricos são mostrados para validar as nossas propostas e a sua aplicação na área de detecção de falhas paramétricas. Todas as funções foram implementadas no MATLAB e conformam um toolbox para identificação de SDP em modelos de regressão / Abstract: This work is about the identification of the dependency among parameters and states in regression models of stochastic time series. The discovery of that dependency can be useful to obtain a more accurate model structure. Conventional recursive algorithms for estimation of Time Variable Parameters do not provide good results in models with state-dependent parameters (SDP) because these may have highly non-linear and even chaotic behavior. This work is based on Peter Young's studies about Off-Line SDP. Young's methods to data entropy reduction are discussed and some data transformations are proposed for this. Later, are proposed some changes on the Off-Line algorithm in order to improve its velocity, accuracy, and tractability to generate the On-Line version. Finally, three numeric examples to validate our proposal are shown. All the functions were implemented in MATLAB and conform a Toolbox to the SDP identification in regression models / Mestrado / Automação / Mestre em Engenharia Elétrica
|
308 |
Identificação de sistema dinâmico em dados de estoque imobiliário / Dynamical system identification in real estate stock dataLuiz Paulo Medina de Lima 24 August 2018 (has links)
Modelos preditivos de mercado são ferramentas importantes para tomadores de decisões no âmbito público e privado. Devido à complexidade dinâmica do mercado imobiliário, composta pela interação de dois submercados distintos (mercado de ativos imobiliários e mercado de consumo de espaço) e pela limitação de dados disponíveis, o estudo analítico de mercados imobiliários requer a modelagem paramétrica de um sistema de equações que os descrevam, seguido pela identificação dos parâmetros deste sistema utilizando dados reais de uma região. Neste trabalho, estudamos o modelo dinâmico de mercado imobiliário proposto por Wheaton (1999), criado a partir do popular modelo de quatro quadrantes de autoria de DiPasquale e Wheaton (1996). Utilizamos técnicas de identificação de sistemas para elaborar um modelo de aprendizado para o estoque imobiliário, e o implementamos em Matlab. Aplicamos o método elaborado em dados simulados, para validá-lo, e então aplicamos o mesmo método, com adaptações, em dados reais do mercado imobiliário canadense. Os resultados obtidos validam o método de identificação de sistema dinâmico quando testado em dados simulados, e corroboram o modelo de Wheaton (1999) como modelo preditivo em dados reais. Ademais, os resultados indicam que um modelo que seja capaz de entender a evolução dinâmica dos parâmetros estáticos do modelo de Wheaton (1999), poderia melhorar os resultados deste como ferramenta preditiva. / Predictive market models are important tools for decision-makers in the public and private spheres. Due to the dynamic complexity of the real estate market, consisting of the interaction of two distinct submarkets (real estate asset market and space consumption market) and the lack of real estate data, the analytical study of real estate markets requires the parametric modeling of a system of equations describing them, followed by the identification of the parameters of this system using real data from a region. In this work, we study the dynamic real estate market model proposed by Wheaton (1999), created from the popular four-quadrant model of DiPasquale e Wheaton (1996). We use system identification techniques to develop a learning model for real estate inventory data, and implement it in Matlab. We apply the method devised in simulated data to validate it, and then apply the same method with adaptations in real data of the Canadian real estate market. The results validate the dynamic system identification method when tested in simulated data, and corroborate the Wheaton (1999) model as a predictive model in real data. In addition, the results indicate that a model that is able to understand the dynamic evolution of the static parameters of the Wheaton (1999) model, could improve its results as a predictive tool.
|
309 |
Video Prediction with Invertible Linear EmbeddingsPottorff, Robert Thomas 01 June 2019 (has links)
Using recently popularized invertible neural network We predict future video frames from complex dynamic scenes. Our invertible linear embedding (ILE) demonstrates successful learning, prediction and latent state inference. In contrast to other approaches, ILE does not use any explicit reconstruction loss or simplistic pixel-space assumptions. Instead, it leverages invertibility to optimize the likelihood of image sequences exactly, albeit indirectly.Experiments and comparisons against state of the art methods over synthetic and natural image sequences demonstrate the robustness of our approach, and a discussion of future work explores the opportunities our method might provide to other fields in which the accurate analysis and forecasting of non-linear dynamic systems is essential.
|
310 |
An Application of Sliding Mode Control to Model-Based Reinforcement LearningParisi, Aaron Thomas 01 September 2019 (has links)
The state-of-art model-free reinforcement learning algorithms can generate admissible controls for complicated systems with no prior knowledge of the system dynamics, so long as sufficient (oftentimes millions) of samples are available from the environ- ment. On the other hand, model-based reinforcement learning approaches seek to leverage known optimal or robust control to reinforcement learning tasks by mod- elling the system dynamics and applying well established control algorithms to the system model. Sliding-mode controllers are robust to system disturbance and modelling errors, and have been widely used for high-order nonlinear system control. This thesis studies the application of sliding mode control to model-based reinforcement learning. Computer simulation results demonstrate that sliding-mode control is viable in the setting of reinforcement learning. While the system performance may suffer from problems such as deviations in state estimation, limitations in the capacity of the system model to express the system dynamics, and the need for many samples to converge, this approach still performs comparably to conventional model-free reinforcement learning methods.
|
Page generated in 0.1234 seconds