A hierarchical machine learning framework for real-time reconstruction of urban wind fields from sparse sensor networks

Keyphrases

• Proper Orthogonal Decomposition 100%
• Unseen 100%
• Reynolds number 100%
• Machine Learning Framework 100%
• Coarse-to-fine 100%
• Real-time Reconstruction 100%
• Ill-conditioned Problem 100%
• Urban Wind Field 100%
• Hierarchical Machine Learning 100%
• Sparse Sensor Network 100%
• Sensing Constraints 100%
• Three-dimensional (3D) 50%
• High-resolution 50%
• Flow Structure 50%
• Wind Tunnel Test 50%
• Large Eddy Simulation 50%
• Single-stage 50%
• Flow Field 50%
• Time-resolved 50%
• Computationally Intensive 50%
• High-fidelity CFD 50%
• Wind Direction 50%
• Wind Field 50%
• Hierarchical Level 50%
• Sensor Networks 50%
• Structural Health Monitoring 50%
• Tractability 50%
• Japan 50%
• Sensor Placement 50%
• Temporal Dynamics 50%
• Reconstruction Accuracy 50%
• Order Reduction 50%
• Digital Twin Application 50%
• Spatial Structure 50%
• Spatial-temporal 50%
• Vertical Plane 50%
• Learning Improvement 50%
• Practical Deployment 50%
• Operational Use 50%
• Staged Approach 50%
• High-resolution Mapping 50%
• Resolution Level 50%
• Sparse Measurements 50%
• Urban Flow 50%
• Decomposition Coefficient 50%
• Multiple Inflows 50%
• Reduced Dimensionality 50%
• Temporal Modeling 50%
• Simulation Dataset 50%
• Reconstruction Fidelity 50%
• Stable Learning 50%
• Half-cylinder 50%
• Robust Generalization 50%
• Constraint-aware 50%
• Inflow Direction 50%
• Coefficient Space 50%
• Hierarchical Reconstruction 50%
• Unsteady Dynamics 50%
• Sparse Information 50%
• Unseen Conditions 50%
• Niigata 50%
• Temporal Segment 50%
• Near Real-time Operation 50%
• Intermediate Resolution 50%
• Multiple Heights 50%
• Urban District 50%
• Structure Dynamics 50%
• Full-field Reconstruction 50%
• Winds Reconstruction 50%
• Order Learning 50%
• Fine Mapping 50%
• Hierarchical Formulation 50%
• Height Level 50%

Engineering

• Learning System 100%
• High Resolution 100%
• Sensor Network 100%
• Proper Orthogonal Decomposition 100%
• Reynolds' Number 100%
• Computational Fluid Dynamics 50%
• Dimensionality 50%
• Flow Structure 50%
• Flow Distribution 50%
• Multiscale 50%
• Field Measurement 50%
• Structural Health Monitoring 50%
• Lstm 50%
• Japan 50%
• Spatial Structure 50%
• Digital Twin 50%
• Wind Tunnels 50%
• Decomposition Coefficient 50%
• Large Eddy Simulation 50%
• Constrainedness 50%

Computer Science

• Machine Learning 100%
• Learning System 100%
• Learning Framework 100%
• Reynolds Number 100%
• Proper Orthogonal Decomposition 100%
• Sensor Network 100%
• Digital Twin 50%
• Long Short-Term Memory Network 50%
• Temporal Modeling 50%
• Hierarchical Level 50%
• Sensor Placement 50%
• Temporal Dynamic 50%
• Spatial Structure 50%
• Reconstruction Accuracy 50%
• Vertical Plane 50%
• Decomposition Coefficient 50%
• Sparse Information 50%
• Resolution Level 50%