Data Analytics for Smart Infrastructure: Asset Management and Network Performance

Cover
Half Title
Title Page
Copyright Page
Contents
Preface
Author
Chapter 1: AI Empowering Infrastructure: The Road to Smartness
	1.1. Introduction
	1.2. Statistics and Machine Learning: Two Sides of the Same Coin
		1.2.1. Agility
		1.2.2. Replicability
		1.2.3. Applicability
		1.2.4. Efficiency
		1.2.5. Capability
	1.3. How Machine Learning Makes Infrastructure Solutions Smarter
		1.3.1. Situation Awareness
		1.3.2. Predictive analysis
		1.3.3. Decisions Support
	1.4. Purpose of This Book
		1.4.1. Audience of the Book
Chapter 2: Asset Anomaly Identification – Damage Detection in Structural Health Monitoring
	2.1. Background
	2.2. Techniques
		2.2.1. Dimensionality Reduction
		2.2.2. Damage Detection
		2.2.3. Damage Localization and Estimation
	2.3. Case Studies
		2.3.1. A Laboratory-Based Building Structure
		2.3.2. The Sydney Harbour Bridge
	2.4. Summary
Chapter 3: Network Performance Evaluation – Delay Propagation on Large Scale Railway Systems
	3.1. Background
	3.2. Methodology
		3.2.1. Problem Formulation and Background
		3.2.2. Preliminary
		3.2.3. Conditional Bayesian Delay Propagation
	3.3. Case Study
		3.3.1. Experimental Setting
		3.3.2. Delay Propagation Prediction from the Temporal Dimensions
		3.3.3. Delay Propagation Prediction from the Spatial Dimensions
		3.3.4. Delay Propagation in Different Scenarios
	3.4. Summary
Chapter 4: Network Status Monitoring – Signal Aspect Detection for Railway Networks
	4.1. Background
	4.2. Preliminary
	4.3. Object Detection Model for Signal Detection
		4.3.1. Model Architecture
		4.3.2. Model Training
	4.4. Image Segmentation Model for Track Detection
		4.4.1. Spatial Path
		4.4.2. Context Path
		4.4.3. Network Architecture
	4.5. Target Signal Detection Model
		4.5.1. Current Track Location
		4.5.2. Target Signal Light Location
	4.6. Summary
Chapter 5: Underground Vessel: Water Pipe Failure Prediction
	5.1. Background
		5.1.1. Pipe Failure Prediction as a Machine Learning Problem
	5.2. Replacement Prediction
		5.2.1. Replacement Prediction as a Machine Learning Problem
		5.2.2. Machine Learning in Survival Analysis
		5.2.3. Predict Deterministic Time-to-Event
		5.2.4. Prediction as Probability
		5.2.5. Predict Binary Survival by Multitask Learning
	5.3. Maintenance Prediction
		5.3.1. Data Understanding and Processing
		5.3.2. Point Process to Model Maintenance Records Data
		5.3.3. Non-Parametric Models for Water Pipe Failure Prediction
	5.4. Summary
Chapter 6: Long-Term Prediction of Water Supply Networks Condition
	6.1. Background
		6.1.1. Long-Term Prediction Clues
	6.2. Machine Learning Frameworks for Long-Term Prediction
		6.2.1. Digitised Rules
		6.2.2. Step-Forwarding Prediction
		6.2.3. Grouping Assets
		6.2.4. Sequence-to-Sequence Prediction
		6.2.5. Conformal Prediction
		6.2.6. Multiple Resolution
	6.3. Long-Term Prediction Models
		6.3.1. Point Process for Long-Term Prediction
		6.3.2. Scale Optimisation
		6.3.3. Adaption Models
		6.3.4. Long-Term Recurrent Neural Network
		6.3.5. Conformal Prediction for Non-Parametric Model
	6.4. Case Study on Water Pipe Failure Prediction with Uncertainty
	6.5. Summary
Chapter 7: Service Demand Prediction – Passenger Flow
	7.1. Background
		7.1.1. IoT Technologies in Railway System of Smart Cities
		7.1.2. Passenger Flow Prediction in ITS
	7.2. Preliminaries
	7.3. Stage 1: Next-Day Passenger Flow Prediction Model
		7.3.1. Decomposing Component for Model Inputs (DC1)
		7.3.2. Influential Factors
		7.3.3. Self-Attention-Based Prediction Component
		7.3.4. Decomposing Component for Passenger Flow Reallocation (DC2)
	7.4. Stage 2: Real-Time Fine-Tuning Model
		7.4.1. Fast Short-Term Fine-Tuning
		7.4.2. Real-Time Adjustment for Interchange and Platform Traffic
	7.5. Case Study
		7.5.1. Experimental Data
		7.5.2. Performance Compassion on Next-Day Passenger Flow Prediction
		7.5.3. Performance Compassion on Real-Time Fine-Tuning Prediction
		7.5.4. Effectiveness of Components
		7.5.5. Fine-Tuning Model on Emergencies
		7.5.6. Public Holiday Scenarios
	7.6. Summary
Chapter 8: Prioritising Risk Assets for Infrastructure Maintenance
	8.1. Background
	8.2. Water Pipe Failure Prediction
		8.2.1. Feature Engineering with Domain Knowledge
		8.2.2. Ensemble Learning
		8.2.3. Feature Importance
	8.3. Group Level Prioritisation
	8.4. Case Study for Water Assets Prioritisation
		8.4.1. Model Validation
		8.4.2. Integration of Consequence
	8.5. Summary
Chapter 9: Adapting Dynamic Behaviour Evolution in Structural Health Monitoring
	9.1. Background
	9.2. A Concept Drift Adaptation Perspective
	9.3. Case Studies
		9.3.1. The Sydney Harbour Bridge
		9.3.2. A Reinforced Concrete Beam
		9.3.3. The Infante D. Henrique Bridge
	9.4. Summary
Chapter 10: Smart Sensing and Preventative Maintenance
	10.1. Background
	10.2. Zones Prioritisation for Sensors Deployment
	10.3. Smart Sensing for Water Networks
		10.3.1. Sensor Deployment and Data Logging
		10.3.2. Machine Learning and Smart Sensing Techniques for Leak Detection
	10.4. Case Study for Water Loss Saving
		10.4.1. Acoustic Monitoring for Leak Detection
		10.4.2. Pipes Prioritisation
		10.4.3. Validation Results
	10.5. Summary
References
Index