A5062002874- Rock Mechanics and Modeling
- Groundwater flow and contamination studies
- Hydraulic Fracturing and Reservoir Analysis
- Drilling and Well Engineering
- Grouting, Rheology, and Soil Mechanics
- Geotechnical Engineering and Analysis
- Dam Engineering and Safety
- Geotechnical Engineering and Underground Structures
- Landslides and related hazards
- Geotechnical and Geomechanical Engineering
- Geomechanics and Mining Engineering
- Tunneling and Rock Mechanics
- Geotechnical Engineering and Soil Stabilization
- Seismic Imaging and Inversion Techniques
- Geophysical Methods and Applications
- Hydrocarbon exploration and reservoir analysis
- Geotechnical Engineering and Soil Mechanics
- Mineral Processing and Grinding
- Reinforcement Learning in Robotics
- CO2 Sequestration and Geologic Interactions
- Geoscience and Mining Technology
- Domain Adaptation and Few-Shot Learning
- Enhanced Oil Recovery Techniques
- Geophysical and Geoelectrical Methods
- Robotic Path Planning Algorithms
Northeastern University
2016-2025
Zhejiang Chinese Medical University
2025
Zhejiang University
2024-2025
Dalian University of Technology
2024
Taiyuan Heavy Industry (China)
2024
Chinese Academy of Sciences
2017-2023
University of Chinese Academy of Sciences
2023
Universidad del Noreste
2022-2023
Aerospace Information Research Institute
2023
Suzhou University of Science and Technology
2022-2023
To optimize flapping foil performance, in the current study we apply deep reinforcement learning (DRL) to plan non-parametric motion, as traditional control techniques and simplified motions cannot fully model nonlinear, unsteady high-dimensional foil–vortex interactions. Therefore, a DRL training framework is proposed based on proximal policy optimization algorithm transformer architecture, where initialized from sinusoidal expert display. We first demonstrate effectiveness of DRL-training...
Fine-tuning large-scale pretrained vision models to downstream tasks is a standard technique for achieving state-of-the-art performance on computer benchmarks. However, fine-tuning the whole model with millions of parameters inefficient as it requires storing same-sized new copy each task. In this work, we propose LoRand, method better tradeoff between task and number trainable parameters. LoRand generates tiny adapter structures low-rank synthesis while keeping original backbone fixed,...
The complicated geological conditions and hazards are challenging problems during tunnel construction, which will cause great losses of life property. Therefore, reliable prediction defective features, such as faults, karst caves groundwater, has important practical significances theoretical values. In this paper, we presented the criteria for detecting typical anomalies using seismic (TSP) method. ground penetrating radar (GPR) signal response to water-bearing structures was used...
Abstract After the excavation of underground engineering, failure and instability surrounding rock under hydro-mechanical coupling conditions is a common type engineering disaster. However, mechanical characteristics have not been fully revealed, suitable models for stability analysis are very scarce. Therefore, series triaxial compression cyclic loading unloading tests were carried out to study characteristics, deformation parameters different confining pressures pore pressures. Then, based...
For fluid flow in fracture networks, the local water head loss will occur due to convergence or deflection at intersections, which obviously affect characteristics of networks. In this study, equations losses for intersections were derived based on work-energy principle. Then, nonuniform analyzed by numerical simulations and influences including velocities, rate distributions patterns studied. Finally, influence total networks was quantitatively. Through a unified relationship between index...