Estimating fuzzy membership functions parameters by the levenberg-marquardt algorithm

In previous papers from the authors fuzzy model identification methods were discussed. The bacterial algorithm for extracting fuzzy rule base from a training set was presented. The Levenberg-Marquardt algorithm was also proposed for determining membership functions in fuzzy systems. In this paper the Levenberg-Marquardt technique is improved to optimise the membership functions in the fuzzy rules without Ruspini-partition. The class of membership functions investigated is the trapezoidal one as it is general enough and widely used. The method can be easily extended to arbitrary piecewise linear functions as well.

Levenberg-Marquardt application to two-phase nonlinear parameter estimation for finned-tube coil evaporators

A procedure for calculation of refrigerant mass flow rate is implemented in the distributed numerical model to simulate the flow in finned-tube coil dry-expansion evaporators, usually found in refrigeration and air-conditioning systems. Two-phase refrigerant flow inside the tubes is assumed to be one-dimensional, unsteady, and homogeneous. In themodel the effects of refrigerant pressure drop and the moisture condensation from the air flowing over the external surface of the tubes are considered. The results obtained are the distributions of refrigerant velocity, temperature and void fraction, tube-wall temperature, air temperature, and absolute humidity. The finite volume method is used to discretize the governing equations. Additionally, given the operation conditions and the geometric parameters, the model allows the calculation of the refrigerant mass flow rate. The value of mass flow rate is computed using the process of parameter estimation with the minimization method of Levenberg-Marquardt minimization. In order to validate the developed model, the obtained results using HFC-134a as a refrigerant are compared with available data from the literature.

Identificação de danos estruturais utilizando dados no domínio do tempo provenientes de ensaios de vibração

O presente trabalho aborda o problema de identificação de danos em uma estrutura a partir de sua resposta impulsiva. No modelo adotado, a integridade estrutural é continuamente descrita por um parâmetro de coesão. Sendo assim, o Modelo de Elementos Finitos (MEF) é utilizado para discretizar tanto o campo de deslocamentos, quanto o campo de coesão. O problema de identificação de danos é, então, definido como um problema de otimização, cujo objetivo é minimizar, em relação a um vetor de parâmetros nodais de coesão, um funcional definido a partir da diferença entre a resposta impulsiva experimental e a correspondente resposta prevista por um MEF da estrutura. A identificação de danos estruturais baseadas no domínio do tempo apresenta como vantagens a aplicabilidade em sistemas lineares e/ou com elevados níveis de amortecimento, além de apresentar uma elevada sensibilidade à presença de pequenos danos. Estudos numéricos foram realizados considerando-se um modelo de viga de Euler-Bernoulli simplesmente apoiada. Para a determinação do posicionamento ótimo do sensor de deslocamento e do número de pontos da resposta impulsiva, a serem utilizados no processo de identificação de danos, foi considerado o Projeto Ótimo de Experimentos. A posição do sensor e o número de pontos foram determinados segundo o critério D-ótimo. Outros critérios complementares foram também analisados. Uma análise da sensibilidade foi realizada com o intuito de identificar as regiões da estrutura onde a resposta é mais sensível à presença de um dano em um estágio inicial. Para a resolução do problema inverso de identificação de danos foram considerados os métodos de otimização Evolução Diferencial e Levenberg-Marquardt. Simulações numéricas, considerando-se dados corrompidos com ruído aditivo, foram realizadas com o intuito de avaliar a potencialidade da metodologia de identificação de danos, assim como a influência da posição do sensor e do número de dados considerados no processo de identificação. Com os resultados obtidos, percebe-se que o Projeto Ótimo de Experimentos é de fundamental importância para a identificação de danos.

Estudos de sorção de um corante aniônico modelo em partículas de quitosana reticulada

Textile activity results in effluents with a variety of dyes. Among the several processes for dye-uptaking from these wastewaters, sorption is one of the most effective methods, chitosan being a very promising alternative for this end. The sorption of Methyl Orange by chitosan crosslinked particles was approached using equilibrium and kinetic analyses at different pH s. Besides the standard pseudo-order analysis normally effectuated (i.e. pseudo-first-order and pseudo-second-order), a novel approach involving a pseudo-nth-order kinetics was used, nbeing determined via non-linear regression, using the Levenberg-Marquardt method. Zeta potential measurements indicated that electrostatic interactions were important for the sorption process. Regarding equilibrium experiments, data were well fitted to a hybrid Langmuir-Freundlich isotherm, and estimated Gibbs free energy of adsorption as a function of mass of dye per area of chitosan showed that the process of adsorption becomes more homogeneous as the pH of the continuous phase decreased. Considering the kinetics of sorption, although a pseudo-nth-order description yielded good fits, a kinetic equation involving diffusion adsorption phenomena was found to be more consistent in terms of a physicochemical description of the sorption process

Um método previsor-corretor primal-dual de pontos interiores barreira logarítmica modificada, com estratégias de convergência global e de ajuste cúbico, para problemas de programação não-linear e não-convexa

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Ein inverses Problem für die degeneriert parabolische Richardsgleichung

In der vorliegenden Arbeit werden zwei physikalischeFließexperimente an Vliesstoffen untersucht, die dazu dienensollen, unbekannte hydraulische Parameter des Materials, wiez. B. die Diffusivitäts- oder Leitfähigkeitsfunktion, ausMeßdaten zu identifizieren. Die physikalische undmathematische Modellierung dieser Experimente führt auf einCauchy-Dirichlet-Problem mit freiem Rand für die degeneriertparabolische Richardsgleichung in derSättigungsformulierung, das sogenannte direkte Problem. Ausder Kenntnis des freien Randes dieses Problems soll dernichtlineare Diffusivitätskoeffizient derDifferentialgleichung rekonstruiert werden. Für diesesinverse Problem stellen wir einOutput-Least-Squares-Funktional auf und verwenden zu dessenMinimierung iterative Regularisierungsverfahren wie dasLevenberg-Marquardt-Verfahren und die IRGN-Methode basierendauf einer Parametrisierung des Koeffizientenraumes durchquadratische B-Splines. Für das direkte Problem beweisen wirunter anderem Existenz und Eindeutigkeit der Lösung desCauchy-Dirichlet-Problems sowie die Existenz des freienRandes. Anschließend führen wir formal die Ableitung desfreien Randes nach dem Koeffizienten, die wir für dasnumerische Rekonstruktionsverfahren benötigen, auf einlinear degeneriert parabolisches Randwertproblem zurück.Wir erläutern die numerische Umsetzung und Implementierungunseres Rekonstruktionsverfahrens und stellen abschließendRekonstruktionsergebnisse bezüglich synthetischer Daten vor.

Numerical analysis of the relation between interactions and structure in a molecular fluid

Coarse graining is a popular technique used in physics to speed up the computer simulation of molecular fluids. An essential part of this technique is a method that solves the inverse problem of determining the interaction potential or its parameters from the given structural data. Due to discrepancies between model and reality, the potential is not unique, such that stability of such method and its convergence to a meaningful solution are issues.rnrnIn this work, we investigate empirically whether coarse graining can be improved by applying the theory of inverse problems from applied mathematics. In particular, we use the singular value analysis to reveal the weak interaction parameters, that have a negligible influence on the structure of the fluid and which cause non-uniqueness of the solution. Further, we apply a regularizing Levenberg-Marquardt method, which is stable against the mentioned discrepancies. Then, we compare it to the existing physical methods - the Iterative Boltzmann Inversion and the Inverse Monte Carlo method, which are fast and well adapted to the problem, but sometimes have convergence problems.rnrnFrom analysis of the Iterative Boltzmann Inversion, we elaborate a meaningful approximation of the structure and use it to derive a modification of the Levenberg-Marquardt method. We engage the latter for reconstruction of the interaction parameters from experimental data for liquid argon and nitrogen. We show that the modified method is stable, convergent and fast. Further, the singular value analysis of the structure and its approximation allows to determine the crucial interaction parameters, that is, to simplify the modeling of interactions. Therefore, our results build a rigorous bridge between the inverse problem from physics and the powerful solution tools from mathematics. rn

Regularization methods for the solution of a nonlinear least-squares problem in tomography

In this work we study a polyenergetic and multimaterial model for the breast image reconstruction in Digital Tomosynthesis, taking into consideration the variety of the materials forming the object and the polyenergetic nature of the X-rays beam. The modelling of the problem leads to the resolution of a high-dimensional nonlinear least-squares problem that, due to its nature of inverse ill-posed problem, needs some kind of regularization. We test two main classes of methods: the Levenberg-Marquardt method (together with the Conjugate Gradient method for the computation of the descent direction) and two limited-memory BFGS-like methods (L-BFGS). We perform some experiments for different values of the regularization parameter (constant or varying at each iteration), tolerances and stop conditions. Finally, we analyse the performance of the several methods comparing relative errors, iterations number, times and the qualities of the reconstructed images.

An inverse optimization strategy to determine single crystal mechanical behavior from polycrystal tests: Application to AZ31 Mg alloy

An inverse optimization strategy was developed to determine the single crystal properties from experimental results of the mechanical behavior of polycrystals. The polycrystal behavior was obtained by means of the finite element simulation of a representative volume element of the microstructure in which the dominant slip and twinning systems were included in the constitutive equation of each grain. The inverse problem was solved by means of the Levenberg-Marquardt method, which provided an excellent fit to the experimental results. The iterative optimization process followed a hierarchical scheme in which simple representative volume elements were initially used, followed by more realistic ones to reach the final optimum solution, leading to important reductions in computer time. The new strategy was applied to identify the initial and saturation critical resolved shear stresses and the hardening modulus of the active slip systems and extension twinning in a textured AZ31 Mg alloy. The results were in general agreement with the data in the literature but also showed some differences. They were partially explained because of the higher accuracy of the new optimization strategy but it was also shown that the number of independent experimental stress-strain curves used as input is critical to reach an accurate solution to the inverse optimization problem. It was concluded that at least three independent stress-strain curves are necessary to determine the single crystal behavior from polycrystal tests in the case of highly textured Mg alloys.

Ajuste del modelo de Drucker Prager extendido para la simulación de la respuesta mecánica de un vertisol mediante el método de elementos finitos

La mecanización de las labores del suelo es la causa, por su consumo energético e impacto directo sobre el medio ambiente, que más afecta a la degradación y pérdida de productividad de los suelos. Entre los factores de disminución de la productividad se deben considerar la compactación, la erosión, el encostramiento y la pérdida de estructura. Todo esto obliga a cuidar el manejo agrícola de los suelos tratando de mejorar las condiciones del suelo y elevar sus rendimientos sin comprometer aspectos económicos, ecológicos y ambientales. En el presente trabajo se adecuan los parámetros constitutivos del modelo de Drucker Prager Extendido (DPE) que definen la fricción y la dilatancia del suelo en la fase de deformación plástica, para minimizar los errores en las predicciones durante la simulación de la respuesta mecánica de un Vertisol mediante el Método de Elementos Finitos. Para lo cual inicialmente se analizaron las bases teóricas que soportan este modelo, se determinaron las propiedades y parámetros físico-mecánicos del suelo requeridos como datos de entrada por el modelo, se determinó la exactitud de este modelo en las predicciones de la respuesta mecánica del suelo, se estimaron mediante el método de aproximación de funciones de Levenberg-Marquardt los parámetros constitutivos que definen la trayectoria de la curva esfuerzo-deformación plástica. Finalmente se comprobó la exactitud de las predicciones a partir de las adecuaciones realizadas al modelo. Los resultados permitieron determinar las propiedades y parámetros del suelo, requeridos como datos de entrada por el modelo, mostrando que su magnitud está en función su estado de humedad y densidad, además se obtuvieron los modelos empíricos de estas relaciones exhibiendo un R2>94%. Se definieron las variables que provocan las inexactitudes del modelo constitutivo (ángulo de fricción y dilatancia), mostrando que las mismas están relacionadas con la etapa de falla y deformación plástica. Finalmente se estimaron los valores óptimos de estos ángulos, disminuyendo los errores en las predicciones del modelo DPE por debajo del 4,35% haciéndelo adecuado para la simulación de la respuesta mecánica del suelo investigado. ABSTRACT The mechanization using farming techniques is one of the main factors that affects the most the soil, causing its degradation and loss of productivity, because of its energy consumption and direct impact on the environment. Compaction, erosion, crusting and loss of structure should be considered among the factors that decrease productivity. All this forces the necessity to take care of the agricultural-land management trying to improve soil conditions and increase yields without compromising economic, ecological and environmental aspects. The present study was aimed to adjust the parameters of the Drucker-Prager Extended Model (DPE), defining friction and dilation of soil in plastic deformation phase, in order to minimize the error of prediction when simulating the mechanical response of a Vertisol through the fine element method. First of all the theoretic fundamentals that withstand the model were analyzed. The properties and physical-mechanical parameters of the soil needed as input data to initialize the model, were established. And the precision of the predictions for the mechanical response of the soil was assessed. Then the constitutive parameters which define the path of the plastic stress-strain curve were estimated through Levenberg-Marquardt method of function approximations. Lastly the accuracy of the predictions from the adequacies made to the model was tested. The results permitted to determine those properties and parameters of the soil, needed in order to initialize the model. It showed that their magnitude is in function of density and humidity. Moreover, the empirical models from these relations were obtained: R2>94%. The variables producing inaccuracies in the constitutive model (angle of repose and dilation) were defined, and there was showed that they are linked with the plastic deformation and rupture point. Finally the optimal values of these angles were established, obtaining thereafter error values for the DPE model under 4, 35%, and making it suitable for the simulation of the mechanical response of the soil under study.

A hybrid training method for B-spline neural networks

Current and past research has brought up new views related to the optimization of neural networks. For a fixed structure, second order methods are seen as the most promising. From previous works we have shown how second order methods are of easy applicability to a neural network. Namely, we have proved how the Levenberg-Marquard possesses not only better convergence but how it can assure the convergence to a local minima. However, as any gradient-based method, the results obtained depend on the startup point. In this work, a reformulated Evolutionary algorithm - the Bacterial Programming for Levenberg-Marquardt is proposed, as an heuristic which can be used to determine the most suitable starting points, therefore achieving, in most cases, the global optimum.

A Gradient-Type Method For Real-Time State Estimation Of Water Distribution Networks

Drinking water distribution networks risk exposure to malicious or accidental contamination. Several levels of responses are conceivable. One of them consists to install a sensor network to monitor the system on real time. Once a contamination has been detected, this is also important to take appropriate counter-measures. In the SMaRT-OnlineWDN project, this relies on modeling to predict both hydraulics and water quality. An online model use makes identification of the contaminant source and simulation of the contaminated area possible. The objective of this paper is to present SMaRT-OnlineWDN experience and research results for hydraulic state estimation with sampling frequency of few minutes. A least squares problem with bound constraints is formulated to adjust demand class coefficient to best fit the observed values at a given time. The criterion is a Huber function to limit the influence of outliers. A Tikhonov regularization is introduced for consideration of prior information on the parameter vector. Then the Levenberg-Marquardt algorithm is applied that use derivative information for limiting the number of iterations. Confidence intervals for the state prediction are also given. The results are presented and discussed on real networks in France and Germany.

Prediction of Ground Water Levels in the Uplands of a Tropical Coastal Riparian Wetland using Artificial Neural Networks

Artificial Neural Networks (ANNs) have been found to be a robust tool to model many non-linear hydrological processes. The present study aims at evaluating the performance of ANN in simulating and predicting ground water levels in the uplands of a tropical coastal riparian wetland. The study involves comparison of two network architectures, Feed Forward Neural Network (FFNN) and Recurrent Neural Network (RNN) trained under five algorithms namely Levenberg Marquardt algorithm, Resilient Back propagation algorithm, BFGS Quasi Newton algorithm, Scaled Conjugate Gradient algorithm, and Fletcher Reeves Conjugate Gradient algorithm by simulating the water levels in a well in the study area. The study is analyzed in two cases-one with four inputs to the networks and two with eight inputs to the networks. The two networks-five algorithms in both the cases are compared to determine the best performing combination that could simulate and predict the process satisfactorily. Ad Hoc (Trial and Error) method is followed in optimizing network structure in all cases. On the whole, it is noticed from the results that the Artificial Neural Networks have simulated and predicted the water levels in the well with fair accuracy. This is evident from low values of Normalized Root Mean Square Error and Relative Root Mean Square Error and high values of Nash-Sutcliffe Efficiency Index and Correlation Coefficient (which are taken as the performance measures to calibrate the networks) calculated after the analysis. On comparison of ground water levels predicted with those at the observation well, FFNN trained with Fletcher Reeves Conjugate Gradient algorithm taken four inputs has outperformed all other combinations.

Problemas inversos em processos difusivos com retenção

Um Estudo para a solução numérica do modelo de difusão com retenção, proposta por Bevilacqua et al. (2011), é apresentado, bem como uma formulação implícita para o problema inverso para a estimativa dos parâmetros envolvidos na formulação matemática do modelo. Através de um estudo minucioso da análise de sensibilidade e do cálculo do coeficiente de correlação de Pearson, são identificadas as chances de se obter sucesso na solução do problema inverso através do método determinístico de Levenberg-Marquardt e dos métodos estocásticos Algoritmo de Colisão de Partículas (Particle Collision Algorithm - PCA) e Evolução Diferencial (Differential Evolution - DE). São apresentados os resultados obtidos através destes três métodos de otimização para três casos de conjunto de parâmetros. Foi observada uma forte correlação entre dois destes três parâmetros, o que dificultou a estimativa simultânea dos mesmos. Porém, foi obtido sucesso nas estimativas individuais de cada parâmetro. Foram obtidos bons resultados para os fatores que multiplicam os termos diferenciais da equação que modela o fenômeno de difusão com retenção.