Showing papers by "Bauhaus University, Weimar published in 2021"

Deep autoencoder based energy method for the bending, vibration, and buckling analysis of Kirchhoff plates with transfer learning

Abstract: In this paper, we present a deep autoencoder based energy method (DAEM) for the bending, vibration and buckling analysis of Kirchhoff plates. The DAEM exploits the higher order continuity of the DAEM and integrates a deep autoencoder and the minimum total potential principle in one framework yielding an unsupervised feature learning method. The DAEM is a specific type of feedforward deep neural network (DNN) and can also serve as function approximator. With robust feature extraction capacity, the DAEM can more efficiently identify patterns behind the whole energy system, such as the field variables, natural frequency and critical buckling load factor studied in this paper. The objective function is to minimize the total potential energy. The DAEM performs unsupervised learning based on generated collocation points inside the physical domain so that the total potential energy is minimized at all points. For the vibration and buckling analysis, the loss function is constructed based on Rayleigh’s principle and the fundamental frequency and the critical buckling load is extracted. A scaled hyperbolic tangent activation function for the underlying mechanical model is presented which meets the continuity requirement and alleviates the gradient vanishing/explosive problems under bending. The DAEM is implemented using Pytorch and the LBFGS optimizer. To further improve the computational efficiency and enhance the generality of this machine learning method, we employ transfer learning. A comprehensive study of the DAEM configuration is performed for several numerical examples with various geometries, load conditions, and boundary conditions.

An efficient optimization approach for designing machine learning models based on genetic algorithm

Abstract: Machine learning (ML) methods have shown powerful performance in different application Nonetheless, designing ML models remains a challenge and requires further research as most procedures adopt a trial and error strategy In this study, we present a methodology to optimize the architecture and the feature configurations of ML models considering a supervised learning process The proposed approach employs genetic algorithm (GA)-based integer-valued optimization for two ML models, namely deep neural networks (DNN) and adaptive neuro-fuzzy inference system (ANFIS) The selected variables in the DNN optimization problems are the number of hidden layers, their number of neurons and their activation function, while the type and the number of membership functions are the design variables in the ANFIS optimization problem The mean squared error (MSE) between the predictions and the target outputs is minimized as the optimization fitness function The proposed scheme is validated through a case study of computational material design We apply the method to predict the fracture energy of polymer/nanoparticles composites (PNCs) with a database gathered from the literature The optimized DNN model shows superior prediction accuracy compared to the classical one-hidden layer network Also, it outperforms ANFIS with significantly lower number of generations in GA The proposed method can be easily extended to optimize similar architecture properties of ML models in various complex systems

A Novel Fractional-Order Multiple-Model Type-3 Fuzzy Control for Nonlinear Systems with Unmodeled Dynamics

Abstract: In this paper, a novel control approach is proposed for a class of uncertain nonlinear system with unmodeled dynamics. Each output of the system is modeled by several first-order dynamic fractional-order fuzzy systems. The best dynamic model is selected at a period of time and the control signal is designed based on this model. The dynamic fractional-order models are based on the special case of general type-2 fuzzy systems which are called interval type-3 fuzzy logic systems (IT3FLSs). The adaptation laws for the consequent parameters of IT3FLSs are derived through stability analysis of the fractional-order systems based on the linear matrix inequality approach. The effectiveness of the proposed scheme is verified by normal simulation on the hyperchaotic Lorenz system with unmodeled dynamics, real-time simulation on the chaotic model of the brushless DC motors using Arduino boards and experimental examination on a heat transfer system with fully unknown dynamics.

A review on application of soft computing techniques for the rapid visual safety evaluation and damage classification of existing buildings

Abstract: Seismic vulnerability assessment of existing buildings is of great concern around the world. Different countries develop various approaches and methodologies to overcome the disastrous effects of earthquakes on the structural parameters of the building and the human losses. There are structures still in service with a high seismic vulnerability, which proposes an urgent need for a screening system's damageability grading system. Rapid urbanization and the proliferation of slums give rise to improper construction practices that make the building stock's reliability ambiguous, including old structures that were constructed either when the seismic codes were not advanced or not enforced by law. Despite having a good knowledge of structural analysis, it is impractical to conduct detailed nonlinear analysis on each building in the target area to define their seismic vulnerability. This indicates the necessity of developing a rapid, reliable, and computationally easy method of seismic vulnerability assessment, more commonly known as Rapid Visual Screening (RVS). This method begins with a walk-down survey by a trained evaluator , and an initial score is assigned to the structure. Further, the vulnerability parameters are defined (predictor variables), and the damage grades are defined. Various methods are then adopted to develop an optimum correlation between the parameters and damage grades. Soft Computing (SC) techniques including probabilistic approaches , meta-heuristics, and Artificial Intelligence (AI) theories such as artificial neural networks , machine learning, fuzzy logic, etc. due to their capabilities in targeting inherent imprecision of phenomena in real-world are among the most important and widely used approaches in this regard. In this paper, a comprehensive literature review of the most commonly used and newly developed innovative methodologies in RVS using powerful SC techniques has been presented to shed light on key factors, strengths, and applications of each SC technique in advancing the RVS field of study.

Nonlocal operator method for dynamic brittle fracture based on an explicit phase field model

Abstract: In this work, we present a nonlocal operator method (NOM) for dynamic fracture exploiting an explicit phase field model. The nonlocal strong forms of the phase field and the associated mechanical model are derived as integral forms by variational principle. The equations are decoupled and solved in time by an explicit scheme employing the Verlet-velocity algorithm for the mechanical field and an adaptive sub-step scheme for the phase field model. The sub-step scheme reduces phase field residual adaptively in a few substeps and thus achieves a rate-independent phase field model. The explicit scheme avoids the calculation of the anisotropic stiffness tensor in the implicit phase field model. One advantage of the NOM is its ease in implementation. The method does not require any shape functions and the associated matrices and vectors are obtained automatically after defining the energy of the system. Hence, the approach can be easily extended to more complex coupled problems. Several numerical examples are presented to demonstrate the performance of the current method.

Optimizing the neural network hyperparameters utilizing genetic algorithm

Abstract: Neural networks (NNs), as one of the most robust and efficient machine learning methods, have been commonly used in solving several problems. However, choosing proper hyperparameters (e.g. the numbers of layers and neurons in each layer) has a significant influence on the accuracy of these methods. Therefore, a considerable number of studies have been carried out to optimize the NN hyperparameters. In this study, the genetic algorithm is applied to NN to find the optimal hyperparameters. Thus, the deep energy method, which contains a deep neural network, is applied first on a Timoshenko beam and a plate with a hole. Subsequently, the numbers of hidden layers, integration points, and neurons in each layer are optimized to reach the highest accuracy to predict the stress distribution through these structures. Thus, applying the proper optimization method on NN leads to significant increase in the NN prediction accuracy after conducting the optimization in various examples.

Wear performance and mechanical properties of MWCNT/HDPE nanocomposites for gearing applications

Abstract: MWCNT/HDPE nanocomposite spur gears prepared in concentrations of 0, 0.5, 1, 1.5, 2, and 2.5% by weight using centrifugal ball milling dispersion method followed by CNC milling. SEM examination revealed a sound dispersion of the MWCNT in the HDPE matrix until 2 wt.%. Nanocomposites exhibited higher decomposition temperature and thermal stability than neat HDPE. Yield strength increased linearly in nanocomposites up to 2 wt.%, and then it saturated. Nanofillers' addition steadily increased the young’s modulus up to a weight fraction of 1.5 %, surging rapidly between 1.5 to 2 wt.%. In contrast, its rate of increase declined between 2 and 2.5 wt.%. The Taber abrasion test showed a reduction in wear loss of nanocomposites. The nanocomposites' toughness increased between 0 (neat HDPE) and 1.5 wt.% of MWCNT but declined at higher concentrations due to transition from ductile to brittle nature. At a torque of 5 N-m, the wear performance increased consistently with the increase in the concentration of nanofillers. At 10 N-m, 2.5 wt.% nanocomposite gears displayed a decline due to the increased brittleness. Satisfactory dispersion and developed interphase fervently contributed to the load transfer and mechanical properties. Nanofillers improved the wear resistance, hardness, and lowered plastic deformation. The gear damage mechanism changed from thermal bending to tooth cracking and deflection in the nanocomposite gears. The 2 wt.% MWCNT/HDPE nanocomposites emerged as potential gearing materials with enhanced hardness, tensile properties, uniform dispersion, thermal stability, wear performance, and reasonable toughness.

ML-EHSAPP: a prototype for machine learning-based earthquake hazard safety assessment of structures by using a smartphone app

Abstract: The recent devastating earthquakes have caused severe physical, social, and financial damage worldwide and indicate that many existing buildings, especially in developing countries, are not designe...

The harm in conflating aging with accessibility

Abstract: "The quest for youth—so futile. Age and wisdom have their graces too."— Jean Luc Picard It is an increasingly global phenomenon that societies promote the notion of youth as the preferred state.a In stark contrast to the "wise elder" of ages past, today old age is assumed to be marked by loss of physical and cognitive ability, diminished relevance, and as we are sadly seeing with the COVID-19 pandemic, devalued humanity.18 In many ways, it is not surprising that such stereotypes are reflected in our technologies: tech companies compete for territory in an already overcrowded youth market; whereas older adults,b if considered users at all, are offered little more than fall alarms, activity monitors, and senior-friendly (often lower functionality) versions of existing tools. Meanwhile, there is a growing trend of workers aging out of the tech industry as early as their mid-40s,17 reflecting the higher value placed on the perspectives of those who represent the default target demographic.

On the influence of build orientation on properties of friction stir welded Al–Si10Mg parts produced by selective laser melting

Abstract: Welding and joining of components processed by additive manufacturing (AM) to other AM as well as conventionally produced components is of high importance for industry as this allows to combine advantages of either technique and to produce large-scale structures, respectively. One of the key influencing factors with respect to weldability and mechanical properties of AM components was found to be the inherent microstructural anisotropy of these components. In present work, the precipitation-hardenable Al–Si10Mg was fabricated in different build orientations using selective laser melting (SLM) and subsequently joined by friction stir welding (FSW) in different combinations. Microstructural analysis showed considerable grain refinement in the friction stir zone, however, pronounced softening occurred in this area. The latter can be mainly attributed to changes in the morphology and size of Si particles. Upon combination of different build orientations a remarkable influence on the tensile strength of FSW joints was seen. Cyclic deformation responses of SLM and FSW samples were examined in depth. Fatigue properties of this alloy in the low-cycle fatigue (LCF) regime imply that SLM samples with the building direction parallel to the loading direction show superior performance under cyclic loading as compared to the other conditions and the FSW joints. From results presented solid process-microstructure-property relationships are drawn.

Propagation and attenuation of Rayleigh and pseudo surface waves in viscoelastic metamaterials

Abstract: The development of seismic metamaterials has attracted much research interest in the past decade. Efforts have been made by using experimental and theoretical approaches to isolate buildings and structures susceptible to elastic surface wave damage. However, most seismic metamaterials were designed without considering the viscoelastic effect that widely exists in nature. In this work, we investigate the propagation and attenuation of the Rayleigh and pseudo surface waves (PSWs) in two types of viscoelastic seismic metamaterials, namely, pillared and inclusion-embedded metamaterials, by analyzing the complex band structures and transmission spectra. The complex band structure developed in this work reveals for the first time the existence of PSWs and their propagation properties in inclusion-embedded metamaterials at the surface. These PSW modes are hidden in the traditional ω(k) technique, therefore showing the usefulness of the complex band structure approach. Introducing viscosity to the substrate of both types of seismic metamaterials will enhance the attenuation of both the Rayleigh wave and PSW. For inclusion-embedded metamaterials, the viscoelastic effect in the soft coating layer can have a specific influence only on the PSW. PSWs show advantages to minimize the relative attenuating effect in general. The results in this work will open up great possibilities for designing and optimizing seismic metamaterials in practice.

Modeling the Producibility of 3D Printing in Polylactic Acid Using Artificial Neural Networks and Fused Filament Fabrication.

Abstract: Polylactic acid (PLA) is a highly applicable material that is used in 3D printers due to some significant features such as its deformation property and affordable cost. For improvement of the end-use quality, it is of significant importance to enhance the quality of fused filament fabrication (FFF)-printed objects in PLA. The purpose of this investigation was to boost toughness and to reduce the production cost of the FFF-printed tensile test samples with the desired part thickness. To remove the need for numerous and idle printing samples, the response surface method (RSM) was used. Statistical analysis was performed to deal with this concern by considering extruder temperature (ET), infill percentage (IP), and layer thickness (LT) as controlled factors. The artificial intelligence method of artificial neural network (ANN) and ANN-genetic algorithm (ANN-GA) were further developed to estimate the toughness, part thickness, and production-cost-dependent variables. Results were evaluated by correlation coefficient and RMSE values. According to the modeling results, ANN-GA as a hybrid machine learning (ML) technique could enhance the accuracy of modeling by about 7.5, 11.5, and 4.5% for toughness, part thickness, and production cost, respectively, in comparison with those for the single ANN method. On the other hand, the optimization results confirm that the optimized specimen is cost-effective and able to comparatively undergo deformation, which enables the usability of printed PLA objects.

Nonlocal strong forms of thin plate, gradient elasticity, magneto-electro-elasticity and phase field fracture by nonlocal operator method

Abstract: The derivation of nonlocal strong forms for many physical problems remains cumbersome in traditional methods. In this paper, we apply the variational principle/weighted residual method based on nonlocal operator method for the derivation of nonlocal forms for elasticity, thin plate, gradient elasticity, electro-magneto-elasticity and phase-field fracture method. The nonlocal governing equations are expressed as an integral form on support and dual-support. The first example shows that the nonlocal elasticity has the same form as dual-horizon non-ordinary state-based peridynamics. The derivation is simple and general and it can convert efficiently many local physical models into their corresponding nonlocal forms. In addition, a criterion based on the instability of the nonlocal gradient is proposed for the fracture modelling in linear elasticity. Several numerical examples are presented to validate nonlocal elasticity and the nonlocal thin plate.

Post-neoliberal housing policy? Disentangling recent reforms in New York, Berlin and Vienna:

Abstract: In cities worldwide, the housing question has returned. As demands and proposals by housing movements have grown bolder, city governments are implementing new policies, ranging from small tweaks to...

The spread of breathing air from wind instruments and singers using schlieren techniques

Abstract: In this article, the spread of breathing air when playing wind instruments and singing was investigated and visualized using two methods: (1) schlieren imaging with a schlieren mirror and (2) background-oriented schlieren (BOS). These methods visualize airflow by visualizing density gradients in transparent media. The playing of professional woodwind and brass instrument players, as well as professional classical trained singers, were investigated to estimate the spread distances of the breathing air. For a better comparison and consistent measurement series, a single high and a single low note as well as an extract of a musical piece were investigated. Additionally, anemometry was used to determine the velocity of the spreading breathing air and the extent to which it was still quantifiable. The results presented in this article show there is no airflow escaping from the instruments, which is transported farther than 1.2 m into the room. However, differences in the various instruments have to be considered to assess properly the spread of the breathing air. The findings discussed below help to estimate the risk of cross-infection for wind instrument players and singers and to develop efficacious safety precautions, which is essential during critical health periods such as the current COVID-19 pandemic.