A5072175564- Model Reduction and Neural Networks
- Fluid Dynamics and Vibration Analysis
- Nuclear Engineering Thermal-Hydraulics
- Magnetic properties of thin films
- Probabilistic and Robust Engineering Design
- Wind and Air Flow Studies
- Meteorological Phenomena and Simulations
- Fluid Dynamics and Turbulent Flows
- Characterization and Applications of Magnetic Nanoparticles
- Advanced Numerical Methods in Computational Mathematics
- Physics of Superconductivity and Magnetism
- Hydraulic and Pneumatic Systems
- Advanced Computational Techniques and Applications
- Theoretical and Computational Physics
- Computational Fluid Dynamics and Aerodynamics
- Enhanced Oil Recovery Techniques
- Hydrocarbon exploration and reservoir analysis
- Reservoir Engineering and Simulation Methods
- Medical Image Segmentation Techniques
- Aerodynamics and Acoustics in Jet Flows
- Nuclear reactor physics and engineering
- Underwater Vehicles and Communication Systems
- Numerical methods for differential equations
- Magnetic Properties and Applications
- Urban Heat Island Mitigation
Tongji University
2014-2024
Guangdong University of Technology
2023
Shanghai Dianji University
2023
Swansea University
2018-2022
Imperial College London
2012-2020
Chinese University of Hong Kong, Shenzhen
2017
KTH Royal Institute of Technology
2017
University of Gothenburg
2017
Chinese University of Hong Kong
2017
University of Hong Kong
2017
Summary This paper presents a novel model reduction method: deep learning reduced order model, which is based on proper orthogonal decomposition and methods. The approach recent technological advancement in the field of artificial neural networks. It has advantage nonlinear system with multiple levels representation predicting data. In this work, training data are obtained from high fidelity solutions at selected time levels. long short‐term memory network used to construct set hypersurfaces...
A novel meta-heuristic algorithm named Egret Swarm Optimization Algorithm (ESOA) is proposed in this paper, which inspired by two egret species' hunting behavior (Great and Snowy Egret). ESOA consists of three primary components: a sit-and-wait strategy, aggressive strategy as well discriminant conditions. The learnable guides the to most probable solution applying pseudo gradient estimator. uses random wandering encirclement mechanisms allow for optimal exploration. model utilized balance...
Summary We present a new non‐intrusive model reduction method for the Navier–Stokes equations. The replaces traditional approach of projecting equations onto reduced space with radial basis function (RBF) multi‐dimensional interpolation. main point this is to construct number interpolation functions using RBF scatter method. are used calculate POD coefficients at each time step from earlier steps. advantage that it does not require modifications source code (which would otherwise be very...
A novel non-intrusive reduced order model (NIROM) for fluid–structure interaction (FSI) has been developed. The is based on proper orthogonal decomposition (POD) and radial basis function (RBF) interpolation method. method independent of the governing equations, therefore, it does not require modifications to source code. This first time that a NIROM was constructed FSI phenomena using POD RBF Another novelty this work implementation under framework an unstructured mesh finite element...
Landslide susceptibility mapping is considered to be a prerequisite for landslide prevention and mitigation. However, delineating the spatial occurrence pattern of remains challenge. This study investigates potential application stacking ensemble learning technique assessment. In particular, support vector machine (SVM), artificial neural network (ANN), logical regression (LR), naive Bayes (NB) were selected as base learners method. The resampling scheme Pearson’s correlation analysis...
Abstract This paper presents a new nonlinear non‐intrusive reduced‐order model (NL‐NIROM) that outperforms traditional proper orthogonal decomposition (POD)‐based reduced order (ROM). improvement is achieved through the use of auto‐encoder (AE) and self‐attention based deep learning methods. The novelty this work it uses stacked (SAE) network to project original high‐dimensional dynamical systems onto low dimensional subspace predict fluid dynamics using an method. introduces reduction...
Magnetic skyrmions are topologically protected nanoscale objects, which promising building blocks for novel magnetic and spintronic devices. Here, we investigate the dynamics of a skyrmion driven by spin wave in nanowire. It is found that (i) first accelerated then decelerated exponentially; (ii) it can turn L-corners with both right left turns; (iii) always turns (right) when number positive (negative) T- Y-junctions. Our results will be basis skyrmionic devices wave.
The effective manipulation of skyrmion motion introduces the technologically relevant possibility skyrmion-based spintronics. Here we show how a magnetic can be captured by radially spatial gradient field. A theoretical model governed Thiele equation is employed to study motion. analytical predictions are compared micromagnetic simulations based on Landau–Lifshitz–Gilbert equation, showing that dynamic behavior skyrmions strongly depends strength fields as well radius and Gilbert damping...
The reconstruction and prediction of full-state flows from sparse data are great scientific engineering significance yet remain challenging, especially in applications where and/or subjected to noise. To this end, study proposes a deep-learning assisted non-intrusive reduced order model (named DCDMD) for high-dimensional flow data. Based on the compressed sensing (CS)-dynamic mode decomposition (DMD), DCDMD is distinguished by two novelties. First, matrix defined overcome strict random...
Abstract The aging of operational reactors leads to increased mechanical vibrations in the reactor interior. vibration in-core sensors near their nominal locations is a new problem for neutronic field reconstruction. Current field-reconstruction methods fail handle spatially moving sensors. In this study, we propose Voronoi tessellation technique combination with convolutional neural networks challenge. Observations from movable were projected onto same global structure using tessellation,...
Summary In this article, we describe a non‐intrusive reduction method for porous media multiphase flows using Smolyak sparse grids. This is the first attempt at applying such an reduced‐order modelling (NIROM) based on grids to flows. The advantage of NIROM resides in that its non‐intrusiveness, which means it does not require modifications source code full model. Another novelty uses construct set hypersurfaces representing reduced‐porous problem. implemented under framework unstructured...
Combined methods of first-principles calculations and Landau-Lifshitz-Gilbert (LLG) macrospin simulations are performed to investigate the coherent magnetization switching in MgO/FePt/Pt(001)-based magnetic tunnel junctions triggered by short pulses electric field through control anisotropy energy (MAE) electrically. First-principles indicate that MAE MgO/FePt/Pt(001) film varies linearly with change field, whereas LLG show could induce in-plane reversal free layer tuning pulse parameters....