Abstract
The response‐only model class selection capability of a novel deep convolutional neural network method is examined herein in a simple, yet effective, manner. Specifically, the responses from a unique degree of freedom along with their class information train and validate a one‐dimensional convolutional neural network. In doing so, the network selects the model class of new and unlabeled signals without the need of the system input information, or full system identification. An optional physics‐based algorithm enhancement is also examined using the Kalman filter to fuse the system response signals using the kinematics constraints of the acceleration and displacement data. Importantly, the method is shown to select the model class in slight signal variations attributed to the damping behavior or hysteresis behavior on both linear and nonlinear dynamic systems, as well as on a 3D building finite element model, providing a powerful tool for structural health monitoring applications.
| Original language | English |
|---|---|
| Pages (from-to) | 784-814 |
| Number of pages | 31 |
| Journal | Earthquake Engineering and Structural Dynamics |
| Volume | 53 |
| Issue number | 2 |
| Early online date | 22 Nov 2023 |
| DOIs | |
| Publication status | Published - 28 Feb 2024 |
Bibliographical note
Publisher Copyright:© 2023 The Authors. Earthquake Engineering & Structural Dynamics published by John Wiley & Sons Ltd.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.Acknowledgements
The author would like to gratefully acknowledge the reviewers for their constructive comments, John T. Katsikadelis forthe discussion on benchmark integro-differential equation problems, and Andrew W. Smyth for the previous insightful
discussions on model class selection and Kalman filtering.
Keywords
- convolutional neural networks (CNN)
- One-dimensional CNN (1D CNN)
- deep learning
- Machine Learning
- Neural Network Classification
- Pattern Recognition
- Physics-enhanced Deep Learning
- Kalman Filter
- Data Fusion
- Stochastic Gradient Descent
- Backpropagation
- Mini-batch Training
- Softmax Classification
- Structural Health Monitoring (SHM)
- Structural Dynamics
- Earthquake Engineering
- Seismic Response Analysis
- Vibration-based Monitoring
- Model Class Selection
- Model Class Assessment
- System Identification
- Nonlinear Dynamic Systems
- Linear Dynamic Systems
- Finite Element Modeling (FEM)
- OpenSees
- 3D Building Model
- Structural Model Updating
- Response-only Signals
- Acceleration Signals
- Displacement Signals
- Velocity Signals
- Dynamic State Estimation
- Unknown Input Systems
- Noise Filtering
- Gaussian White Noise
- Time-series Data
- Signal Classification
- Bayesian Model Selection
- Kullback–Leibler Divergence
- Markov Chain Monte Carlo (MCMC)
- Transitional MCMC
- Dynamic Model Averaging
- Unscented Kalman Filter
- Kinematics-based Fusion
- Integro-differential Equations
- Damping Kernel Functions
- Bouc–Wen Model
- Rayleigh Damping
- Real-time Application
- Low-data Training
- Unique DOF Measurement
- Feature Extraction
- Global Average Pooling
- Batch Normalization
- ReLU Activation
- Epoch Optimization
- Sensitivity Analysis
- Training Accuracy
- Loss Minimization
- Long Short-Term Memory (LSTM)
- Gated Recurrent Unit (GRU)
- Recurrent Neural Networks (RNN)
- Clustering Techniques
- Uncertainty Quantification
- Bayesian Latent Force Models
- Extrapolation Capability
- Robustness to Structural Changes
ASJC Scopus subject areas
- Chemical Engineering(all)
- Computer Science(all)
- Energy(all)
- Economics, Econometrics and Finance(all)
- Engineering(all)
- Materials Science(all)
- Mathematics(all)
- Physics and Astronomy(all)
- Decision Sciences(all)
- Environmental Science(all)
