A5100723435- Vibration Control and Rheological Fluids
- Vibration and Dynamic Analysis
- Innovative Energy Harvesting Technologies
- Gamma-ray bursts and supernovae
- Structural Engineering and Vibration Analysis
- Energy Harvesting in Wireless Networks
- Natural Language Processing Techniques
- Pulsars and Gravitational Waves Research
- Topic Modeling
- Railway Engineering and Dynamics
- Astro and Planetary Science
- Advanced Surface Polishing Techniques
- Laser Material Processing Techniques
- Neutrino Physics Research
- Model Reduction and Neural Networks
- Composite Structure Analysis and Optimization
- Fluid Dynamics and Vibration Analysis
- Multimodal Machine Learning Applications
- Seismic Performance and Analysis
- Biomimetic flight and propulsion mechanisms
- Fractional Differential Equations Solutions
- Stellar, planetary, and galactic studies
- Wood Treatment and Properties
- Aeroelasticity and Vibration Control
- Advanced machining processes and optimization
Chinese Academy of Sciences
1992-2024
Tamkang University
2015-2024
Changchun Institute of Optics, Fine Mechanics and Physics
2024
State Key Laboratory of Applied Optics
2024
Changchun University of Technology
2024
North China University of Science and Technology
2023
University of Science and Technology of China
2020-2021
Harbin Institute of Technology
2020-2021
Anhui Institute of Optics and Fine Mechanics
2020-2021
Hefei Institutes of Physical Science
2021
Abstract. Cloud detection and cloud properties have substantial applications in weather forecast, signal attenuation analysis, other cloud-related fields. image segmentation is the fundamental important step deriving cover. However, traditional methods rely on low-level visual features of clouds often fail to achieve satisfactory performance. Deep convolutional neural networks (CNNs) can extract high-level feature information objects achieved remarkable success many On this basis, a novel...
In this paper, we explore the possibility of leveraging Residual Networks (ResNet), a powerful structure in constructing extremely deep neural network for image understanding, to improve recurrent networks (RNN) modeling sequential data.We show that sequence classification tasks, incorporating residual connections into structures yields similar accuracy Long Short Term Memory (LSTM) RNN with much fewer model parameters.In addition, propose two novel models which combine best both learning...
While monolingual data has been shown to be useful in improving bilingual neural machine translation (NMT), effectively and efficiently leveraging for Multilingual NMT (MNMT) systems is a less explored area. In this work, we propose multi-task learning (MTL) framework that jointly trains the model with task on bitext two denoising tasks data. We conduct extensive empirical studies MNMT 10 language pairs from WMT datasets. show proposed approach can improve quality both high-resource...
Abstract Most protostars have luminosities that are fainter than expected from steady accretion over the protostellar lifetime. The solution to this problem may lie in episodic mass accretion—prolonged periods of very low punctuated by short bursts rapid accretion. However, timescale and amplitude for variability at phase is almost entirely unconstrained. In A James Clerk Maxwell Telescope/SCUBA-2 Transient Survey Protostars Nearby Star-forming Regions, we monitoring monthly with SCUBA-2...
While very deep neural networks have shown effectiveness for computer vision and text classification applications, how to increase the network depth of machine translation (NMT) models better quality remains a challenging problem. Directly stacking more blocks NMT model results in no improvement even drop performance. In this work, we propose an effective two-stage approach with three specially designed components construct deeper models, which result significant improvements over strong...
The formation and evolution of clouds are associated with their thermodynamical microphysical progress. Previous studies have been conducted to collect images using ground-based cloud observation equipment provide important characteristics information. However, most this cannot perform continuous observations during the day night, field view (FOV) is also limited. To address these issues, work proposes a night detection approach integrated into self-made thermal-infrared (TIR) all-sky-view...
Ensemble learning, which aggregates multiple diverse models for inference, is a common practice to improve the accuracy of machine learning tasks. However, it has been observed that conventional ensemble methods only bring marginal improvement neural translation (NMT) when individual are strong or there large number models. In this paper, we study how effectively aggregate NMT under transductive setting where source sentences test set known. We propose simple yet effective approach named...
Deep learning technology has been widely used in various field recent years. This study intends to use deep algorithms analyze the aeroelastic phenomenon and compare differences between Neural Network (DNN) Long Short-term Memory (LSTM) applied on flutter speed prediction. In this present work, DNN LSTM are address complex systems by superimposing multi-layer Artificial Network. Under such an architecture, neurons neural network can extract features from flight data. Instead of...
Scratches on optical components induce laser damage and limit the increase in power. Magnetorheological finishing (MRF) is a highly deterministic manufacturing technology that can improve surface roughness of components. Although MRF has exhibited significant potential for reducing subsurface removing scratches, principle mechanism behind scratch removal are not sufficiently understood. In this study, theory fluid mechanics used to analyze pressure, velocity, particle trajectory distribution...
This study investigated the effect of a pendulum tuned mass damper (PTMD) on vibration slender two-dimensional (2D) rigid body with 1:2 internal resonance. Focus is placed damping various parameters PTMD preventing resonance system. The instruments used include fixed points plots, time response and Poincaré maps, which were compared for confirmation accuracy. Lagrange's equation employed to derive equations motion method multiple scales (MOMS) applied analyzing this nonlinear model....
This study examined the vibrations of a hinged-free nonlinear beam placed on elastic foundation. The authors found that specific combinations an modulus in foundation resulted 1:3 internal resonances first and second modes beam. prompted adding dynamic vibration absorber (DVA) order to prevent resonance suppress vibrations. When DVA was at free end beam, boundaries were time dependent. Thus, shifting polynomial function used convert nonhomogeneous boundary conditions into homogeneous...
This study investigates the application of deep learning models—specifically Deep Neural Networks (DNN), Long Short-Term Memory (LSTM), and (LSTM-NN)—to predict panel flutter in aerospace structures. The goal is to improve accuracy efficiency predicting aeroelastic behaviors under various flight conditions. Utilizing a supersonic flat plate as main structure, research integrates conditions into equation. resulting structural vibration data create large-scale database for training models....
This study considered a slender hinged-free beam with suspension cables simulated using nonlinear springs. We added time-dependent boundary dynamic vibration absorber (DVA) that was suspended at the free end of to reduce and prevent internal resonance. DVA various spring constants were optimal mass range for in main structure also proposed.
This research proposes an energy harvesting system that collects the downward airflow from a helicopter or multi-axis unmanned rotary-wing aircraft and uses this wind force to drive magnet rotate, generating repulsive force, which causes double elastic steel slap each other vibrate periodically in order generate more electricity than traditional system. The design concept of vibration mechanism study is allow carrying another piezoelectric patch form set (DES VEH) systems. theoretical DES...
This study examines how the vibration absorbers influence stability of nonlinear flow-solid interaction systems. A novel approach is proposed for analysis dynamic two-dimensional system using internal resonance contour plot (IRCP) and flutter speed (FSCP). The considered a planar rigid-body with plunge pitch vibrations. two ends body are supported by cubic springs quadratic damping. absorber attached beneath also as rigid body, mass position adjusted optimization reduction. method multiple...
This study uses machine learning to predict the convergence results of Duffing equation with and without damping. The represents a nonlinear second-order differential interesting behavior in undamped free vibration forced Convergence alternates randomly between 1 -1 vibration, depending on initial conditions. For damping, multiple factors influence patterns. We utilize fourth-order Runge-Kutta method collect for both Machine techniques, specifically long short term memory (LSTM) LSTM-Neural...
This study aims to enhance conventional vibration energy harvesting systems (VEHs) by repositioning the piezoelectric patch (PZT) in middle of a fixed-fixed elastic steel sheet instead root, as is commonly case. The system subjected an axial simple harmonic force at one end induce transversal and deformation. To further improve power conversion, baffle strategically installed point maximum deflection, introducing slapping augment electrical harvesting. Employing theory nonlinear beams,...
Abstract The objective of this study was to determine and predict the withdrawal resistance or pull-out load common wire nails embedded in radial, tangential cross-sectional grain orientation Douglas fir (Presudotsuga menziesii) sugar maple (Acer sacharum) samples. Four lead-hole diameters 1.5, 2.0, 2.5 3.0 mm were used create various interference fits. Nails with a diameter 3.38 driven into samples depth 10 for experiments. overall found be lower than that Strength values each sample...