Cover Image

Interpretable machine learning for the analysis, design, assessment, and informed decision making for civil infrastructure /

Interpretable Machine Learning for the Analysis, Design, Assessment, and Informed Decision Making for Civil Infrastructure highlights the growing trend of fostering machine learning to realize contemporary, smart, and safe infrastructure. This volume delves into the latest advancements in machine le...

Full description

Saved in:
Bibliographic Details
Other Authors: Naser, M. Z. (Editor)
Format: Electronic eBook
Language:English
Published: Cambridge, MA : Woodhead Publishing, [2024]
Series:Woodhead Publishing series in civil and structural engineering.
Subjects:
Machine learning > Industrial applications.
Civil engineering > Data processing.
Apprentissage automatique > Applications industrielles.
Civil engineering > Data processing
Machine learning > Industrial applications
Online Access:ScienceDirect
Tags: Add Tag
No Tags, Be the first to tag this record!
Table of Contents:
  • Front Cover
  • Interpretable Machine Learning for the Analysis, Design, Assessment, and Informed Decision Making for Civil Infrastructure
  • Copyright
  • Contents
  • List of contributors
  • Preface
  • 1 Integrated schematic design method for shear wall structures: a practical application of generative adversarial networks
  • 1.1 Introduction
  • 1.2 Literature review
  • 1.2.1 Learning-based structural design method
  • 1.2.2 Parametric modeling
  • 1.2.3 Transformation between pixel image and structured design data
  • 1.3 Framework
  • 1.4 Preprocessing of architectural CAD drawings
  • 1.5 Intelligent structural design based on GANs
  • 1.5.1 Physics-enhanced GAN
  • 1.5.2 Dataset
  • 1.5.3 Cloud design platform
  • 1.6 Establishment of the structural analysis model
  • 1.6.1 Heterogeneous data transformation
  • 1.6.2 Parametric modeling
  • 1.7 Case study
  • 1.7.1 Evaluation method of shear wall layout
  • 1.7.2 Basic information of a typical case
  • 1.7.3 Detailed analyses of a typical case
  • 1.8 Conclusions
  • Acknowledgments
  • Appendix 1.A Python code to extract shear wall elements from a pixel image
  • References
  • 2 Leveraging machine learning techniques to support a holistic performance-based seismic design of civil structures
  • 2.1 Introduction
  • 2.2 Background
  • 2.3 ML-assisted holistic seismic design
  • 2.3.1 Objectives and motivation
  • 2.3.2 Framework overview
  • 2.3.2.1 Early design module
  • 2.3.2.2 Detailed design module
  • 2.3.3 Performance data inventories
  • 2.3.3.1 Vulnerability data
  • 2.3.3.2 Environmental performance data
  • 2.3.3.3 Seismic recovery data
  • 2.4 Case study
  • 2.4.1 Design problem definition
  • 2.4.2 Building inventory description
  • 2.4.2.1 Finite element modeling
  • 2.4.2.2 GM selection and time-history analysis
  • 2.4.2.3 Seismic loss assessment
  • 2.4.2.4 Environmental impact assessment.
  • 2.4.3 ML-assisted early design
  • 2.4.3.1 Data-driven surrogate models
  • 2.4.3.2 Sequential design space exploration
  • 2.4.4 ML-assisted detailed design
  • 2.5 Conclusion
  • 2.5.1 Summary
  • 2.5.2 Limitations
  • Appendix 2.A Example codes for ML implementation
  • 2.A.1 Example SQL query from INSSEPT
  • 2.A.2 Example python code to train an ML algorithm
  • References
  • 3 Deep learning methods for concrete structure damage inspection
  • 3.1 Introduction
  • 3.2 Main focus of the chapter
  • 3.2.1 Recognition and measurement of PC components using point cloud
  • 3.2.1.1 Data preprocessing
  • 3.2.1.2 Rebar, concrete, and sleeve recognition
  • 3.2.1.3 Boundary and corner estimation
  • 3.2.1.4 Position and dimension estimation
  • 3.2.2 Image-based local damage recognition
  • 3.2.2.1 Active learning framework with difficulty learning mechanism
  • 3.2.2.2 Architecture of crack-FPN
  • 3.2.3 Anomaly detection in structures using vibration response analysis
  • 3.2.3.1 Three-channel input establishment and multistage sampling
  • 3.2.3.2 GANomaly
  • 3.2.3.3 CNN
  • 3.2.4 Image and point cloud-based finite element analysis
  • 3.2.4.1 Damage detection
  • 3.2.4.2 ICP-based DLT registration
  • 3.2.4.3 FE model generation
  • 3.3 Conclusions
  • References
  • 4 Explainable computational intelligence method to evaluate the damage on concrete surfaces compared to traditional visual i...
  • 4.1 Introduction
  • 4.2 Traditional visual inspection
  • 4.3 Neural networks
  • 4.3.1 Artificial neural networks
  • 4.3.2 Elements of artificial neural networks
  • 4.3.3 Weighted sum function and step activation function
  • 4.3.4 How computers see images
  • 4.3.5 Convolutional neural networks
  • 4.3.6 Convolution filters
  • 4.3.7 How to apply filters to learn features
  • 4.3.8 Pooling layers
  • 4.3.9 Fully connected layers
  • 4.4 Image classification problem
  • 4.4.1 Dataset.
  • 4.4.1.1 Training and testing dataset
  • 4.4.2 Performance metrics
  • 4.4.3 Computational details
  • 4.4.4 Generating the classifier
  • 4.4.4.1 Building the baseline model
  • 4.4.4.2 Proposing a robust CNN architecture
  • 4.5 Conclusions
  • Acknowledgment
  • References
  • 5 Smart building fire safety design driven by artificial intelligence
  • Abbreviations
  • 5.1 Introduction
  • 5.2 AI model and numerical fire database
  • 5.2.1 Early-stage fire and detection
  • 5.2.2 Spatiotemporal prediction of the tenability profile inside the atrium
  • 5.3 Intelligent Fire Engineering Tool (IFETool)
  • 5.3.1 Prediction on the response of ceiling-mounted sprinkler/detector
  • 5.3.2 Tenability assessment for large-space atrium
  • 5.4 Perspectives of AI-driven building fire safety design
  • 5.5 Summary
  • Acknowledgment
  • References
  • 6 The potential of deep learning in dynamic maintenance scheduling for thermal energy storage chiller plants
  • 6.1 Background
  • 6.2 Overview of deep learning
  • 6.2.1 Machine learning types
  • 6.2.2 Types of deep learning applications for predictive maintenance
  • 6.2.2.1 Convolutional neural networks (CNNs)
  • 6.2.2.2 Recurrent neural networks (RNNs)
  • 6.2.2.3 Multilayer perceptrons (MLPs)
  • 6.2.2.4 Autoencoders
  • 6.2.2.5 Restricted Boltzmann machines (RBMs)
  • 6.2.2.6 Generative adversarial networks (GANs)
  • 6.2.2.7 Radial basis function networks (RBFNs)
  • 6.2.2.8 Deep belief networks (DBNs)
  • 6.2.2.9 Self-organizing maps (SOMs)
  • 6.3 Deep learning for thermal-energy-storage chiller plants
  • 6.4 Proposed intelligent maintenance system for TES chiller
  • 6.5 Potential of deep learning-based maintenance scheduling
  • 6.5.1 Limitations and challenges
  • 6.5.2 Future directions
  • 6.6 Summary
  • References
  • 7 Role of intelligent data analysis to enhance GPR data interoperability: road transports
  • 7.1 Introduction.
  • 7.1.1 Background knowledge
  • 7.1.2 Objectives and main contribution
  • 7.2 Fundamentals of GPR images and complexity in data interpretations
  • 7.3 Intelligent data analysis approaches for GPR applications
  • 7.4 Real case study using IDA with GPR data for road transports
  • 7.4.1 Layer interfaces identification using Attention UNet model: experimental results
  • 7.4.2 Methodology
  • 7.4.3 GPR data preparation
  • 7.4.4 Results and conclusion
  • 7.5 NDT indicators for road assessment condition performance
  • 7.6 Final remarks and future prospective
  • Acknowledgments
  • References
  • 8 AI for large-scale evacuation modeling: promises and challenges
  • 8.1 Introduction
  • 8.2 Literature review
  • 8.2.1 Planning models
  • 8.2.1.1 Statistical models
  • 8.2.1.2 Machine learning models
  • 8.2.2 Real-time models
  • 8.2.2.1 Statistical models
  • 8.2.2.2 Deep learning models
  • 8.3 Conceptual framework
  • 8.3.1 Data needs
  • 8.3.2 Suitable AI algorithms
  • 8.3.2.1 AI-based planning models
  • Decision tree
  • Random forest
  • Support vector machine
  • Bayesian network
  • K-nearest neighbor
  • K-means
  • Hierarchical clustering
  • Principal component analysis
  • Deep reinforcement learning
  • 8.3.2.2 AI-based real-time models
  • Long short-term memory neural network
  • Bidirectional encoder representations from transformers
  • Situational-aware multigraph convolutional recurrent network
  • 8.3.3 Model training mechanism
  • 8.3.4 Performance metrics
  • 8.3.5 Model explainability
  • 8.4 Discussion
  • 8.4.1 Major challenges
  • 8.4.2 Future research directions
  • 8.5 Conclusions
  • Acknowledgments
  • Appendix 8.A
  • References
  • 9 On application of machine learning classifiers in evaluating liquefaction potential of civil infrastructure
  • 9.1 Introduction
  • 9.2 Overview of the used machine learning algorithms
  • 9.2.1 Logistic regression.
  • 9.2.2 Random forest classifier
  • 9.2.3 SVM classifier
  • 9.2.4 KNN classifier
  • 9.2.5 Na̐ve Bayes classifier
  • 9.3 Explanatory data analysis
  • 9.4 Model development process
  • 9.4.1 Feature selection
  • 9.4.2 Development of classifiers
  • 9.4.3 Calibration methods
  • 9.4.3.1 Assessing calibration
  • 9.5 Performance of the developed models
  • 9.6 Risk-based liquefication evaluation
  • 9.6.1 Optimal decision
  • 9.7 Illustrative example
  • 9.8 Conclusions
  • Data availability statement
  • References
  • 10 Explainable machine learning model for prediction of axial capacity of strengthened CFST columns
  • 10.1 Introduction
  • 10.2 Experimental datasets
  • 10.3 FE simulation
  • 10.3.1 General
  • 10.3.2 Material modeling
  • 10.3.2.1 Concrete material
  • 10.3.2.2 Steel and CFRP materials
  • 10.3.3 FE model verification
  • 10.3.4 Parametric study
  • 10.4 Machine learning
  • 10.4.1 General
  • 10.4.2 ANN
  • 10.4.3 XGBoost algorithm
  • 10.4.4 SHAP
  • 10.4.5 Results and discussion
  • 10.4.5.1 ANN model
  • 10.4.5.2 XGBoost
  • 10.4.5.2.1 Model validation
  • 10.4.5.2.2 SHAP plots
  • 10.5 Strength reduction factors
  • 10.6 Conclusions
  • Appendix 10.A Supplementary data
  • References
  • 11 Harnessing data from benchmark testing for the development of spalling detection techniques using deep learning
  • 11.1 Introduction
  • 11.2 Bridge specimen
  • 11.2.1 Specimen background
  • 11.2.2 Used earthquake excitations
  • 11.2.3 Spalling damage estimation summary
  • 11.3 Proposed convolutional neural networks
  • 11.3.1 Proposed CNN models
  • 11.3.2 Proposed techniques
  • 11.4 Convolutional neural network training and testing
  • 11.4.1 CNN training and testing for different models
  • 11.4.2 CNN training and testing for different augmentation operations
  • 11.4.3 CNN training and testing for different hyperparameters
  • 11.4.4 Proposed architecture.

Similar Items