A5061544769- Adhesion, Friction, and Surface Interactions
- Probabilistic and Robust Engineering Design
- Force Microscopy Techniques and Applications
- Additive Manufacturing Materials and Processes
- Meteorological Phenomena and Simulations
- Climate variability and models
- Additive Manufacturing and 3D Printing Technologies
- Statistical Methods and Inference
- Surface Roughness and Optical Measurements
- Target Tracking and Data Fusion in Sensor Networks
- Advanced Statistical Methods and Models
- Structural Health Monitoring Techniques
- Welding Techniques and Residual Stresses
- Hydrology and Drought Analysis
- Manufacturing Process and Optimization
- Time Series Analysis and Forecasting
- Mechanical stress and fatigue analysis
- Diamond and Carbon-based Materials Research
- Solidification and crystal growth phenomena
- GNSS positioning and interference
- Gear and Bearing Dynamics Analysis
- Bayesian Modeling and Causal Inference
- Advanced Multi-Objective Optimization Algorithms
- Fluid Dynamics and Turbulent Flows
- Cancer Mechanisms and Therapy
RWTH Aachen University
2020-2025
Hanoi University of Science and Technology
2021
Technische Universität Braunschweig
2018-2021
University of Liège
2015-2018
Defence Science and Technology Group
1998
In this study, a data-driven deep learning model for fast and accurate prediction of temperature evolution melting pool size metallic additive manufacturing processes are developed. The study focuses on bulk experiments the M4 high-speed steel material powder manufactured by Direct Energy Deposition. Under non-optimized process parameters, many deposited layers (above 30) generate large changes microstructure through sample depth caused high sensitivity cladding thermal history. A 2D finite...
Stiction is a failure mode of microelectromechanical systems (MEMS) involving permanent adhesion moving surfaces. Models stiction typically describe the as multiple asperity adhesive contact between random rough surfaces, and they thus require sufficiently accurate statistical representation surface, which may be non-Gaussian. If caused primarily by in only small portion apparent area contacting number contacts asperities not statistically significant for homogenized model to representative....
In this contribution, several case studies with data uncertainties are presented which have been performed in individual projects as part of the DFG (German Research Foundation) Priority Programme SPP 1886 “Polymorphic uncertainty modelling for numerical design structures.” all models derived from engineering problems describing concepts handling and incorporating measurement data, either model input parameters or system response. The first study deals polymorphic uncertain based on computer...
This work deals with parameter identification problems in which uncertainties are modeled using random sets (RS), i.e., set-valued variables. Dempster's rule of combination is applied for replacing the role Bayes' to infer posterior, also a RS. The considered framework allows accounting mixed epistemic-aleatory uncertainty descriptions such as probability boxes and intervals. In this paper, we aim at an efficient computational method sample posterior RS stochastic methods developed Bayesian...
This paper presents the machine learning-based ensemble conditional mean filter (ML-EnCMF) — a filtering method based on (CMF) previously introduced in literature. The updated of CMF matches that posterior, obtained by applying Bayes' rule filter's forecast distribution. Moreover, we show CMF's covariance coincides with expected covariance. Implementing EnCMF requires computing (CM). A likelihood-based estimator is prone to significant errors for small sizes, causing divergence. We develop...
This work studies the uncertainties of adhesive contact problems for reduced size structures, e.g. stiction failure microelectromechanical systems (MEMS). In MEMS, because large surface to volume ratio, surfaces forces, such as van der Waals forces and capillary are dominant in comparison with body forces. As these force magnitudes strongly depend on distance, when two contacting rough, distances vary, physical areas limited at highest asperities surfaces. Therefore, between rough can suffer...
This paper compares parameterization techniques used for modeling the vertical profile of ionosphere. In particular, comparisons three‐layer parameterizations driven by ionosonde data are performed. Quasi‐parabolic, Chapman, and polynomial models ionosphere investigated. Optimization applied at varying stages sensor output, ranging from direct inversion digital ionogram image to fitting true height output estimation parameters, described. A Kalman‐based filter is optionally track parameters....
Abstract This paper considers Bayesian identification of macroscopic bone material characteristics given digital image correlation (DIC) data. As the evaluation full posterior distribution is known to be computationally intense, here we consider approximate estimation in a Newton‐like manner by using theory conditional expectation. The approach extended include epistemic uncertainties process modelling prior.
This study quantifies the effects of uncertainty raised from process parameters, material properties, and boundary conditions in directed energy deposition (DED) M4 High-Speed Steel using deep learning (DL)-based probabilistic approach. A DL-based surrogate model is first constructed data obtained a finite element (FE) model, which was validated against experiment. Then, sources are characterized by method propagated Monte-Carlo (MC) method. Lastly, sensitivity analysis (SA) variance-based...
Abstract Uncertainty of random variables is commonly characterized from measurement data. In practice, data might be insufficient in order to obtain an accurate probability model. this work, we assume that the type distribution considered variable known a priori, and use hierarchical parametric box (p‐box) – which set distributions whose parameters are uncertain account for limited Using Bayes' rule, knowledge about variability these updated. Propagation uncertainties through computational...
<p>Filtering is an uncertainty quantification technique that refers to the inference of states dynamical systems from noisy observations. This work proposes a machine learning-based filtering method for tracking high-dimensional non-Gaussian state-space models with non-linear dynamics and sparse Our filter based on conditional expectation mean uses machine-learning techniques approximate (CM). The contribution this twofolds: (i) we demonstrate theoretically assimilated...
A random set is a generalisation of variable, i.e. set-valued variable. The theory allows unification other uncertainty descriptions such as interval mass belief function in Dempster-Shafer evidence, possibility theory, and probability distributions. aim this work to develop non-deterministic inference framework, including approximation sampling method, that deals with the inverse problems which represented using sets. proposed method yields posterior based on intersection prior measurement...
Most of the gold mining plants in Ecuador are south country.Gold extraction processes use large amounts cyanide and produce industrial wastewater that is discharged into rivers.These discharges loaded with cyanide, which very reactive easily forms bonds a wide variety chemical elements heavy metals forming different species.This study aims to simulate speciation effluents from small-scale Southern analyze these species as potential water pollutants.Visual MINTEQ simulated models temperature,...
Abstract This article presents a novel deep learning‐based ensemble conditional mean filter (DL‐EnCMF) for nonlinear data assimilation. The filter's key component is the approximation of expectation (CE) using neural networks (DNNs). We implement DL‐EnCMF tracking states Lorenz‐63 system. Numerical results show that outperforms Kalman (EnKF)—a common technique