A5089094092- Fluid Dynamics and Heat Transfer
- Surface Modification and Superhydrophobicity
- Combustion and flame dynamics
- Plant Surface Properties and Treatments
- Advanced Combustion Engine Technologies
- Heat transfer and supercritical fluids
- Catalytic Processes in Materials Science
- Natural Language Processing Techniques
- Topic Modeling
- Nanomaterials for catalytic reactions
- Catalysis and Oxidation Reactions
- Catalytic C–H Functionalization Methods
- Radical Photochemical Reactions
- Sulfur-Based Synthesis Techniques
- Fluid Dynamics Simulations and Interactions
- Advanced Photocatalysis Techniques
- Catalysis and Hydrodesulfurization Studies
- Phase Equilibria and Thermodynamics
- Veterinary medicine and infectious diseases
- Speech Recognition and Synthesis
- Advanced Text Analysis Techniques
- Microwave-Assisted Synthesis and Applications
- COVID-19 diagnosis using AI
- Environmental remediation with nanomaterials
- Domain Adaptation and Few-Shot Learning
Wuhan Polytechnic University
2023-2024
National Institute of Clean and Low-Carbon Energy
2024
Tsinghua University
2020-2024
Sheng Jing Hospital
2023
Shenyang Pharmaceutical University
2022
Kyushu University
2020-2022
Zhejiang University of Technology
2021-2022
Baidu (China)
2020-2021
Jiangsu Normal University
2019
Wuhan University of Technology
2018
Pre-trained models have achieved state-of-the-art results in various Natural Language Processing (NLP) tasks. Recent works such as T5 and GPT-3 shown that scaling up pre-trained language can improve their generalization abilities. Particularly, the model with 175 billion parameters shows its strong task-agnostic zero-shot/few-shot learning capabilities. Despite success, these large-scale are trained on plain texts without introducing knowledge linguistic world knowledge. In addition, most an...
While ridged, spherical, or cone superhydrophobic surfaces have been extensively utilized to explore the droplet impact dynamics and possibility of reducing contact time, with a single small pillar received less attention. Here, we report rebound splashing phenomena droplets on various single-pillar pillars having smaller equal sizes compared droplets. Our results indicate that inhibit flat ones, former sequentially exhibit three morphologies top, bottom, breakup rebounds increasing Weber...
The impact behavior of a water droplet on small cylindrical superhydrophobic targets is studied numerically and theoretically. A numerical model using the volume fluid method developed to simulate process targets. verified by comparing calculated results with experimental observations in our previous work reference. influences Weber number target-to-droplet diameter ratio (less than one) behaviors, including profile deformation factor, are investigated. indicate that larger accelerates...
Droplet impacts on solid surfaces are ubiquitous in nature and industry. Before impact, the droplet shape may be affected by gravity, shear flow, electric magnetic fields, inducing non-spherical droplets. However, most previous studies focused impact dynamics of spherical In this study, we conduct experiments, simulations, theoretical analyses to investigate behaviors ellipsoidal water droplets whose symmetry axis is perpendicular surface. particular, explore maximum spreading energy...
SiYu Ding, Junyuan Shang, Shuohuan Wang, Yu Sun, Hao Tian, Hua Wu, Haifeng Wang. Proceedings of the 59th Annual Meeting Association for Computational Linguistics and 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.
Pre-trained language models have achieved state-of-the-art results in various Natural Language Processing (NLP) tasks. GPT-3 has shown that scaling up pre-trained can further exploit their enormous potential. A unified framework named ERNIE 3.0 was recently proposed for pre-training large-scale knowledge enhanced and trained a model with 10 billion parameters. outperformed the on NLP In order to explore performance of 3.0, we train hundred-billion-parameter called Titan 260 parameters...
Simple and smart directional transport of impacting droplets is great significance for future life industry, but there are still many challenges due to complex hydrodynamics. Inspired by Setaria viridis leaves, we propose an efficient strategy manipulate the split based on simple design liquid-repellent surfaces decorated with a ridge, report large-scale, high-speed, fast-response bidirectional droplets. Upon wirelike droplet splits into two parts, which have considerable horizontal velocity...
Deep learning-based surrogate models have received wide attention for efficient and cost-effective predictions of fluid flows combustion, while their hyperparameter settings often lack generalizable guidelines. This study examines two different types models, convolutional autoencoder (CAE)-based reduced order (ROMs) fully connected (FCAE)-based ROMs, emulating hydrogen-enriched combustion from a triple-coaxial nozzle jet. The performances these ROMs are discussed in detail, with an emphasis...
In advanced aeropropulsion engines, liquid fuel is often injected into the combustor at supercritical pressures, where flow dynamics are distinct from subcritical counterpart. Large-eddy simulation, combined with real-fluid thermodynamics and transport processes of a N-dodecane jet in oxygen crossflow, presented different pressures jet-to-crossflow momentum flux ratios ([Formula: see text]). Various vortical structures discussed detail. The results show that, same velocity ratio 0.75,...
High-fidelity simulations of mixing and combustion processes are computationally demanding time-consuming, hindering their wide application in industrial design optimization. This study proposes projection-based reduced order models (ROMs) to predict spatial distributions physical fields for multi-species problems a fast accurate manner. The developed ROMs explore the suitability various regression methods, including kriging, multivariate polynomial (MPR), k-nearest neighbors (KNN), deep...
The selective hydrogenolysis of the C–O bonds lignin-derived aryl ethers into aromatics is challenging because it always accompanied by hydrogenation (HYD). Most metal-supported catalysts tested so far exhibit high efficiency for bond cleavage diphenyl ether (DPE, 4-O-5 linkage in lignin) but low selectivity toward valuable aromatic compounds. Here, we report our discovery showing feasibility controlling products tuning catalyst support and reaction conditions. Pt/γ-Al2O3 exhibited a higher...
In advanced aero-engines, kerosene is often transversely injected into the combustor at supercritical pressure, where shorter jet penetration depth may result in poor mixing and local hot spots near wall. Elevating nozzle proposed to remedy these issues, flowfield complexity increases as a of intricate interactions among jet, crossflow, stack wake. The distinct flow dynamics elevated dodecane jets crossflow (EJICF) pressure are numerically investigated using large eddy simulation. effects...
The energy conversion efficiency (the ratio of the maximum jumping kinetic to surface released from droplet coalescence) is an essential indicator self-propelled droplets, which determines its value for applications in various fields. In practical condensation process, initial states multidroplets with different sizes and distributions have a significant effect on efficiency, but mechanism behind this remains unclear. This paper reveals droplets multidroplet (mainly three droplets)...