MICHAEL G. PECHT, PHD, is Chair Professor in Mechanical Engineering and Professor in Applied Mathematics, Statistics and Scientific Computation at the University of Maryland, USA. He is the Founder and Director of the Center for Advanced Life Cycle Engineering (CALCE) at the University of Maryland, USA, which is funded by more than 150 leading electronics companies. Dr. Pecht is an IEEE, ASME, SAE, and IMAPS Fellow and serves as editor-in-chief of IEEE Access. He has written more than 30 books, 700 technical articles, and has 8 patents. MYEONGSU KANG, PHD, is currently a Research Associate at the Center for Advanced Life Cycle Engineering (CALCE), University of Maryland, USA. His expertise is in data analytics, machine learning, system modeling, and statistics for prognostics and systems health management. He has authored/coauthored more than 60 publications in leading journals and conference proceedings.

• List of Contributors xxiiiPreface xxviiAbout the Contributors xxxvAcknowledgment xlviiList of Abbreviations xlix1 Introduction to PHM 1Michael G. Pecht andMyeongsu Kang1.1 Reliability and Prognostics 11.2 PHM for Electronics 31.3 PHM Approaches 61.3.1 PoF-Based Approach 61.3.1.1 Failure Modes, Mechanisms, and Effects Analysis (FMMEA) 71.3.1.2 Life-Cycle Load Monitoring 81.3.1.3 Data Reduction and Load Feature Extraction 101.3.1.4 Data Assessment and Remaining Life Calculation 121.3.1.5 Uncertainty Implementation and Assessment 131.3.2 Canaries 141.3.3 Data-Driven Approach 161.3.3.1 Monitoring and Reasoning of Failure Precursors 161.3.3.2 Data Analytics and Machine Learning 201.3.4 Fusion Approach 231.4 Implementation of PHM in a System of Systems 241.5 PHM in the Internet ofThings (IoT) Era 261.5.1 IoT-Enabled PHM Applications: Manufacturing 271.5.2 IoT-Enabled PHM Applications: Energy Generation 271.5.3 IoT-Enabled PHM Applications: Transportation and Logistics 281.5.4 IoT-Enabled PHM Applications: Automobiles 281.5.5 IoT-Enabled PHM Applications: Medical Consumer Products 291.5.6 IoT-Enabled PHM Applications:Warranty Services 291.5.7 IoT-Enabled PHM Applications: Robotics 301.6 Summary 30References 302 Sensor Systems for PHM 39Hyunseok Oh,Michael H. Azarian, Shunfeng Cheng, andMichael G. Pecht2.1 Sensor and Sensing Principles 392.1.1 Thermal Sensors 402.1.2 Electrical Sensors 412.1.3 Mechanical Sensors 422.1.4 Chemical Sensors 422.1.5 Humidity Sensors 442.1.6 Biosensors 442.1.7 Optical Sensors 452.1.8 Magnetic Sensors 452.2 Sensor Systems for PHM 462.2.1 Parameters to be Monitored 472.2.2 Sensor System Performance 482.2.3 Physical Attributes of Sensor Systems 482.2.4 Functional Attributes of Sensor Systems 492.2.4.1 Onboard Power and Power Management 492.2.4.2 Onboard Memory and Memory Management 502.2.4.3 Programmable SamplingMode and Sampling Rate 512.2.4.4 Signal Processing Software 512.2.4.5 Fast and Convenient Data Transmission 522.2.5 Reliability 532.2.6 Availability 532.2.7 Cost 542.3 Sensor Selection 542.4 Examples of Sensor Systems for PHM Implementation 542.5 Emerging Trends in Sensor Technology for PHM 59References 603 Physics-of-Failure Approach to PHM 61Shunfeng Cheng, Nagarajan Raghavan, Jie Gu, Sony Mathew, and Michael G. Pecht3.1 PoF-Based PHM Methodology 613.2 Hardware Configuration 623.3 Loads 633.4 Failure Modes, Mechanisms, and Effects Analysis (FMMEA) 643.4.1 Examples of FMMEA for Electronic Devices 683.5 Stress Analysis 713.6 Reliability Assessment and Remaining-Life Predictions 733.7 Outputs from PoF-Based PHM 773.8 Caution and Concerns in the Use of PoF-Based PHM 783.9 Combining PoF with Data-Driven Prognosis 80References 814 Machine Learning: Fundamentals 85Myeongsu Kang and Noel Jordan Jameson4.1 Types of Machine Learning 854.1.1 Supervised, Unsupervised, Semi-Supervised, and Reinforcement Learning 864.1.2 Batch and Online Learning 884.1.3 Instance-Based and Model-Based Learning 894.2 Probability Theory in Machine Learning: Fundamentals 904.2.1 Probability Space and Random Variables 914.2.2 Distributions, Joint Distributions, and Marginal Distributions 914.2.3 Conditional Distributions 914.2.4 Independence 924.2.5 Chain Rule and Bayes Rule 924.3 Probability Mass Function and Probability Density Function 934.3.1 Probability Mass Function 934.3.2 Probability Density Function 934.4 Mean, Variance, and Covariance Estimation 944.4.1 Mean 944.4.2 Variance 944.4.3 Robust Covariance Estimation 954.5 Probability Distributions 964.5.1 Bernoulli Distribution 964.5.2 Normal Distribution 964.5.3 Uniform Distribution 974.6 Maximum Likelihood and Maximum A Posteriori Estimation 974.6.1 Maximum Likelihood Estimation 974.6.2 Maximum A Posteriori Estimation 984.7 Correlation and Causation 994.8 Kernel Trick 1004.9 Performance Metrics 1024.9.1 Diagnostic Metrics 1024.9.2 Prognostic Metrics 105References 1075 Machine Learning: Data Pre-processing 111Myeongsu Kang and Jing Tian5.1 Data Cleaning 1115.1.1 Missing Data Handling 1115.1.1.1 Single-Value Imputation Methods 1135.1.1.2 Model-Based Methods 1135.2 Feature Scaling 1145.3 Feature Engineering 1165.3.1 Feature Extraction 1165.3.1.1 PCA and Kernel PCA 1165.3.1.2 LDA and Kernel LDA 1185.3.1.3 Isomap 1195.3.1.4 Self-Organizing Map (SOM) 1205.3.2 Feature Selection 1215.3.2.1 Feature Selection: FilterMethods 1225.3.2.2 Feature Selection:WrapperMethods 1245.3.2.3 Feature Selection: Embedded Methods 1245.3.2.4 Advanced Feature Selection 1255.4 Imbalanced Data Handling 1255.4.1 SamplingMethods for Imbalanced Learning 1265.4.1.1 Synthetic Minority Oversampling Technique 1265.4.1.2 Adaptive Synthetic Sampling 1265.4.1.3 Effect of SamplingMethods for Diagnosis 127References 1296 Machine Learning: Anomaly Detection 131Myeongsu Kang6.1 Introduction 1316.2 Types of Anomalies 1336.2.1 Point Anomalies 1346.2.2 Contextual Anomalies 1346.2.3 Collective Anomalies 1356.3 Distance-Based Methods 1366.3.1 MD Calculation Using an Inverse Matrix Method 1376.3.2 MD Calculation Using a Gram–Schmidt Orthogonalization Method 1376.3.3 Decision Rules 1386.3.3.1 Gamma Distribution:Threshold Selection 1386.3.3.2 Weibull Distribution:Threshold Selection 1396.3.3.3 Box-Cox Transformation:Threshold Selection 1396.4 Clustering-Based Methods 1406.4.1 k-Means Clustering 1416.4.2 Fuzzy c-Means Clustering 1426.4.3 Self-Organizing Maps (SOMs) 1426.5 Classification-Based Methods 1446.5.1 One-Class Classification 1456.5.1.1 One-Class Support Vector Machines 1456.5.1.2 k-Nearest Neighbors 1486.5.2 Multi-Class Classification 1496.5.2.1 Multi-Class Support Vector Machines 1496.5.2.2 Neural Networks 1516.6 StatisticalMethods 1536.6.1 Sequential Probability Ratio Test 1546.6.2 Correlation Analysis 1566.7 Anomaly Detection with No System Health Profile 1566.8 Challenges in Anomaly Detection 158References 1597 Machine Learning: Diagnostics and Prognostics 163Myeongsu Kang7.1 Overview of Diagnosis and Prognosis 1637.2 Techniques for Diagnostics 1657.2.1 Supervised Machine Learning Algorithms 1657.2.1.1 Naïve Bayes 1657.2.1.2 Decision Trees 1677.2.2 Ensemble Learning 1697.2.2.1 Bagging 1707.2.2.2 Boosting: AdaBoost 1717.2.3 Deep Learning 1727.2.3.1 Supervised Learning: Deep Residual Networks 1737.2.3.2 Effect of Feature Learning-Powered Diagnosis 1767.3 Techniques for Prognostics 1787.3.1 Regression Analysis 1787.3.1.1 Linear Regression 1787.3.1.2 Polynomial Regression 1807.3.1.3 Ridge Regression 1817.3.1.4 LASSO Regression 1827.3.1.5 Elastic Net Regression 1837.3.1.6 k-Nearest Neighbors Regression 1837.3.1.7 Support Vector Regression 1847.3.2 Particle Filtering 1857.3.2.1 Fundamentals of Particle Filtering 1867.3.2.2 Resampling Methods – A Review 187References 1898 Uncertainty Representation, Quantification, and Management in Prognostics 193Shankar Sankararaman8.1 Introduction 1938.2 Sources of Uncertainty in PHM 1968.3 Formal Treatment of Uncertainty in PHM 1998.3.1 Problem 1: Uncertainty Representation and Interpretation 1998.3.2 Problem 2: Uncertainty Quantification 1998.3.3 Problem 3: Uncertainty Propagation 2008.3.4 Problem 4: Uncertainty Management 2008.4 Uncertainty Representation and Interpretation 2008.4.1 Physical Probabilities and Testing-Based Prediction 2018.4.1.1 Physical Probability 2018.4.1.2 Testing-Based Life Prediction 2018.4.1.3 Confidence Intervals 2028.4.2 Subjective Probabilities and Condition-Based Prognostics 2028.4.2.1 Subjective Probability 2028.4.2.2 Subjective Probabilities in Condition-Based Prognostics 2038.4.3 Why is RUL Prediction Uncertain? 2038.5 Uncertainty Quantification and Propagation for RUL Prediction 2038.5.1 Computational Framework for Uncertainty Quantification 2048.5.1.1 Present State Estimation 2048.5.1.2 Future State Prediction 2058.5.1.3 RUL Computation 2058.5.2 RUL Prediction: An Uncertainty Propagation Problem 2068.5.3 Uncertainty PropagationMethods 2068.5.3.1 Sampling-Based Methods 2078.5.3.2 AnalyticalMethods 2098.5.3.3 Hybrid Methods 2098.5.3.4 Summary of Methods 2098.6 Uncertainty Management 2108.7 Case Study: Uncertainty Quantification in the Power System of an Unmanned Aerial Vehicle 2118.7.1 Description of the Model 2118.7.2 Sources of Uncertainty 2128.7.3 Results: Constant Amplitude Loading Conditions 2138.7.4 Results: Variable Amplitude Loading Conditions 2148.7.5 Discussion 2148.8 Existing Challenges 2158.8.1 Timely Predictions 2158.8.2 Uncertainty Characterization 2168.8.3 Uncertainty Propagation 2168.8.4 Capturing Distribution Properties 2168.8.5 Accuracy 2168.8.6 Uncertainty Bounds 2168.8.7 Deterministic Calculations 2168.9 Summary 217References 2179 PHM Cost and Return on Investment 221Peter Sandborn, ChrisWilkinson, Kiri Lee Sharon, Taoufik Jazouli, and Roozbeh Bakhshi9.1 Return on Investment 2219.1.1 PHM ROI Analyses 2229.1.2 Financial Costs 2249.2 PHM Cost-Modeling Terminology and Definitions 2259.3 PHM Implementation Costs 2269.3.1 Nonrecurring Costs 2269.3.2 Recurring Costs 2279.3.3 Infrastructure Costs 2289.3.4 Nonmonetary Considerations and Maintenance Culture 2289.4 Cost Avoidance 2299.4.1 Maintenance Planning Cost Avoidance 2319.4.2 Discrete-Event Simulation Maintenance PlanningModel 2329.4.3 Fixed-Schedule Maintenance Interval 2339.4.4 Data-Driven (Precursor to Failure Monitoring) Methods 2339.4.5 Model-Based (LRU-Independent)Methods 2349.4.6 Discrete-Event Simulation Implementation Details 2369.4.7 Operational Profile 2379.5 Example PHM Cost Analysis 2389.5.1 Single-Socket Model Results 2399.5.2 Multiple-Socket Model Results 2419.6 Example Business Case Construction: Analysis for ROI 2469.7 Summary 255References 25510 Valuation and Optimization of PHM-Enabled Maintenance Decisions 261Xin Lei, Amir Reza Kashani-Pour, Peter Sandborn, and Taoufik Jazouli10.1 Valuation and Optimization of PHM-Enabled Maintenance Decisions for an Individual System 26210.1.1 A PHM-Enabled Predictive Maintenance OptimizationModel for an Individual System 26310.1.2 Case Study: Optimization of PHM-Enabled Maintenance Decisions for an Individual System (Wind Turbine) 26510.2 Availability 26810.2.1 The Business of Availability: Outcome-Based Contracts 26910.2.2 Incorporating Contract Terms into Maintenance Decisions 27010.2.3 Case Study: Optimization of PHM-Enabled Maintenance Decisions for Systems (Wind Farm) 27010.3 Future Directions 27210.3.1 Design for Availability 27210.3.2 Prognostics-BasedWarranties 27510.3.3 Contract Engineering 276References 27711 Health and Remaining Useful Life Estimation of Electronic Circuits 279Arvind Sai Sarathi Vasan and Michael G. Pecht11.1 Introduction 27911.2 RelatedWork 28111.2.1 Component-Centric Approach 28111.2.2 Circuit-Centric Approach 28211.3 Electronic Circuit Health Estimation Through Kernel Learning 28511.3.1 Kernel-Based Learning 28511.3.2 Health Estimation Method 28611.3.2.1 Likelihood-Based Function for Model Selection 28811.3.2.2 Optimization Approach for Model Selection 28911.3.3 Implementation Results 29211.3.3.1 Bandpass Filter Circuit 29311.3.3.2 DC–DC Buck Converter System 30011.4 RUL Prediction Using Model-Based Filtering 30611.4.1 Prognostics Problem Formulation 30611.4.2 Circuit DegradationModeling 30711.4.3 Model-Based Prognostic Methodology 31011.4.4 Implementation Results 31311.4.4.1 Low-Pass Filter Circuit 31311.4.4.2 Voltage Feedback Circuit 31511.4.4.3 Source of RUL Prediction Error 32011.4.4.4 Effect of First-Principles-Based Modeling 32011.5 Summary 322References 32412 PHM-Based Qualification of Electronics 329Preeti S. Chauhan12.1 Why is Product Qualification Important? 32912.2 Considerations for Product Qualification 33112.3 Review of Current Qualification Methodologies 33412.3.1 Standards-Based Qualification 33412.3.2 Knowledge-Based or PoF-Based Qualification 33712.3.3 Prognostics and Health Management-Based Qualification 34012.3.3.1 Data-Driven Techniques 34012.3.3.2 Fusion Prognostics 34312.4 Summary 345References 34613 PHM of Li-ion Batteries 349Saurabh Saxena, Yinjiao Xing, andMichael G. Pecht13.1 Introduction 34913.2 State of Charge Estimation 35113.2.1 SOC Estimation Case Study I 35213.2.1.1 NN Model 35313.2.1.2 Training and Testing Data 35413.2.1.3 Determination of the NN Structure 35513.2.1.4 Training and Testing Results 35613.2.1.5 Application of Unscented Kalman Filter 35713.2.2 SOC Estimation Case Study II 35713.2.2.1 OCV–SOC-T Test 35813.2.2.2 Battery Modeling and Parameter Identification 35913.2.2.3 OCV–SOC-T Table for Model Improvement 36013.2.2.4 Validation of the Proposed Model 36213.2.2.5 Algorithm Implementation for Online Estimation 36213.3 State of Health Estimation and Prognostics 36513.3.1 Case Study for Li-ion Battery Prognostics 36613.3.1.1 Capacity DegradationModel 36613.3.1.2 Uncertainties in Battery Prognostics 36813.3.1.3 Model Updating via Bayesian Monte Carlo 36813.3.1.4 SOH Prognostics and RUL Estimation 36913.3.1.5 Prognostic Results 37113.4 Summary 371References 37214 PHM of Light-Emitting Diodes 377Moon-Hwan Chang, Jiajie Fan, Cheng Qian, and Bo Sun14.1 Introduction 37714.2 Review of PHM Methodologies for LEDs 37814.2.1 Overview of Available Prognostic Methods 37814.2.2 Data-DrivenMethods 37914.2.2.1 Statistical Regression 37914.2.2.2 Static Bayesian Network 38114.2.2.3 Kalman Filtering 38214.2.2.4 Particle Filtering 38314.2.2.5 Artificial Neural Network 38414.2.3 Physics-Based Methods 38514.2.4 LED System-Level Prognostics 38714.3 Simulation-Based Modeling and Failure Analysis for LEDs 38814.3.1 LED Chip-LevelModeling and Failure Analysis 38914.3.1.1 Electro-optical Simulation of LED Chip 38914.3.1.2 LED Chip-Level Failure Analysis 39314.3.2 LED Package-Level Modeling and Failure Analysis 39514.3.2.1 Thermal and Optical Simulation for Phosphor-Converted White LED Package 39514.3.2.2 LED Package-Level Failure Analysis 39714.3.3 LED System-LevelModeling and Failure Analysis 39914.4 Return-on-Investment Analysis of Applying Health Monitoring to LED Lighting Systems 40114.4.1 ROI Methodology 40314.4.2 ROI Analysis of Applying System Health Monitoring to LED Lighting Systems 40614.4.2.1 Failure Rates and Distributions for ROI Simulation 40714.4.2.2 Determination of Prognostics Distance 41014.4.2.3 IPHM, CPHM, and Cu Evaluation 41214.4.2.4 ROI Evaluation 41714.5 Summary 419References 42015 PHM in Healthcare 431Mary Capelli-Schellpfeffer,Myeongsu Kang, andMichael G. Pecht15.1 Healthcare in the United States 43115.2 Considerations in Healthcare 43215.2.1 Clinical Consideration in ImplantableMedical Devices 43215.2.2 Considerations in Care Bots 43315.3 Benefits of PHM 43815.3.1 Safety Increase 43915.3.2 Operational Reliability Improvement 44015.3.3 Mission Availability Increase 44015.3.4 System’s Service Life Extension 44115.3.5 Maintenance Effectiveness Increase 44115.4 PHM of ImplantableMedical Devices 44215.5 PHM of Care Bots 44415.6 Canary-Based Prognostics of Healthcare Devices 44515.7 Summary 447References 44716 PHM of Subsea Cables 451David Flynn, Christopher Bailey, Pushpa Rajaguru,Wenshuo Tang, and Chunyan Yin16.1 Subsea Cable Market 45116.2 Subsea Cables 45216.3 Cable Failures 45416.3.1 Internal Failures 45516.3.2 Early-Stage Failures 45516.3.3 External Failures 45516.3.4 Environmental Conditions 45516.3.5 Third-Party Damage 45616.4 State-of-the-Art Monitoring 45716.5 Qualifying and Maintaining Subsea Cables 45816.5.1 Qualifying Subsea Cables 45816.5.2 Mechanical Tests 45816.5.3 Maintaining Subsea Cables 45916.6 Data-Gathering Techniques 46016.7 Measuring theWear Behavior of Cable Materials 46116.8 Predicting Cable Movement 46316.8.1 Sliding Distance Derivation 46316.8.2 Scouring Depth Calculations 46516.9 Predicting Cable Degradation 46616.9.1 Volume Loss due to Abrasion 46616.9.2 Volume Loss due to Corrosion 46616.10 Predicting Remaining Useful Life 46816.11 Case Study 47116.12 Future Challenges 47116.12.1 Data-Driven Approach for Random Failures 47116.12.2 Model-Driven Approach for Environmental Failures 47316.12.2.1 Fusion-Based PHM 47316.12.2.2 Sensing Techniques 47416.13 Summary 474References 47517 Connected Vehicle Diagnostics and Prognostics 479Yilu Zhang and Xinyu Du17.1 Introduction 47917.2 Design of an Automatic Field Data Analyzer 48117.2.1 Data Collection Subsystem 48217.2.2 Information Abstraction Subsystem 48217.2.3 Root Cause Analysis Subsystem 48217.2.3.1 Feature-Ranking Module 48217.2.3.2 Relevant Feature Set Selection 48417.2.3.3 Results Interpretation 48617.3 Case Study: CVDP for Vehicle Batteries 48617.3.1 Brief Background of Vehicle Batteries 48617.3.2 Applying AFDA for Vehicle Batteries 48817.3.3 Experimental Results 489Contents xvii17.3.3.1 Information Abstraction 49017.3.3.2 Feature Ranking 49017.3.3.3 Interpretation of Results 49517.4 Summary 498References 49918 The Role of PHM at Commercial Airlines 503RhondaWalthall and Ravi Rajamani18.1 Evolution of Aviation Maintenance 50318.2 Stakeholder Expectations for PHM 50618.2.1 Passenger Expectations 50618.2.2 Airline/Operator/Owner Expectations 50718.2.3 Airframe Manufacturer Expectations 50918.2.4 Engine Manufacturer Expectations 51018.2.5 System and Component Supplier Expectations 51118.2.6 MRO Organization Expectations 51218.3 PHM Implementation 51318.3.1 SATAA 51318.4 PHM Applications 51718.4.1 Engine Health Management (EHM) 51718.4.1.1 History of EHM 51818.4.1.2 EHM Infrastructure 51918.4.1.3 Technologies Associated with EHM 52018.4.1.4 The Future 52318.4.2 Auxiliary Power Unit (APU) Health Management 52418.4.3 Environmental Control System (ECS) and Air Distribution Health Monitoring 52518.4.4 Landing System Health Monitoring 52618.4.5 Liquid Cooling System Health Monitoring 52618.4.6 Nitrogen Generation System (NGS) Health Monitoring 52718.4.7 Fuel Consumption Monitoring 52718.4.8 Flight Control Actuation Health Monitoring 52818.4.9 Electric Power System Health Monitoring 52918.4.10 Structural Health Monitoring (SHM) 52918.4.11 Battery Health Management 53118.5 Summary 532References 53319 PHM Software for Electronics 535Noel Jordan Jameson,Myeongsu Kang, and Jing Tian19.1 PHM Software: CALCE Simulation Assisted Reliability Assessment 53519.2 PHM Software: Data-Driven 54019.2.1 Data Flow 54119.2.2 Master Options 54219.2.3 Data Pre-processing 54319.2.4 Feature Discovery 54519.2.5 Anomaly Detection 54619.2.6 Diagnostics/Classification 54819.2.7 Prognostics/Modeling 55219.2.8 Challenges in Data-Driven PHM Software Development 55419.3 Summary 55720 eMaintenance 559Ramin Karim, Phillip Tretten, and Uday Kumar20.1 From Reactive to Proactive Maintenance 55920.2 The Onset of eMaintenance 56020.3 MaintenanceManagement System 56120.3.1 Life-cycle Management 56220.3.2 eMaintenance Architecture 56420.4 Sensor Systems 56420.4.1 Sensor Technology for PHM 56520.5 Data Analysis 56520.6 Predictive Maintenance 56620.7 Maintenance Analytics 56720.7.1 Maintenance Descriptive Analytics 56820.7.2 Maintenance Analytics and eMaintenance 56820.7.3 Maintenance Analytics and Big Data 56820.8 Knowledge Discovery 57020.9 Integrated Knowledge Discovery 57120.10 User Interface for Decision Support 57220.11 Applications of eMaintenance 57220.11.1 eMaintenance in Railways 57220.11.1.1 Railway Cloud: Swedish Railway Data 57320.11.1.2 Railway Cloud: Service Architecture 57320.11.1.3 Railway Cloud: Usage Scenario 57420.11.2 eMaintenance in Manufacturing 57420.11.3 MEMS Sensors for Bearing Vibration Measurement 57620.11.4 Wireless Sensors for Temperature Measurement 57620.11.5 Monitoring Systems 57620.11.6 eMaintenance Cloud and Servers 57820.11.7 Dashboard Managers 58020.11.8 Alarm Servers 58020.11.9 Cloud Services 58120.11.10 Graphic User Interfaces 58320.12 Internet Technology and Optimizing Technology 585References 58621 Predictive Maintenance in the IoT Era 589Rashmi B. Shetty21.1 Background 58921.1.1 Challenges of a Maintenance Program 59021.1.2 Evolution of Maintenance Paradigms 59021.1.3 Preventive Versus Predictive Maintenance 59221.1.4 P–F Curve 59221.1.5 Bathtub Curve 59421.2 Benefits of a Predictive Maintenance Program 59521.3 Prognostic Model Selection for Predictive Maintenance 59621.4 Internet ofThings 59821.4.1 Industrial IoT 59821.5 Predictive Maintenance Based on IoT 59921.6 Predictive Maintenance Usage Cases 60021.7 Machine Learning Techniques for Data-Driven Predictive Maintenance 60021.7.1 Supervised Learning 60221.7.2 Unsupervised Learning 60221.7.3 Anomaly Detection 60221.7.4 Multi-class and Binary Classification Models 60321.7.5 Regression Models 60421.7.6 Survival Models 60421.8 Best Practices 60421.8.1 Define Business Problem and QuantitativeMetrics 60521.8.2 Identify Assets and Data Sources 60521.8.3 Data Acquisition and Transformation 60621.8.4 Build Models 60721.8.5 Model Selection 60721.8.6 Predict Outcomes and Transform into Process Insights 60821.8.7 Operationalize and Deploy 60921.8.8 Continuous Monitoring 60921.9 Challenges in a Successful Predictive Maintenance Program 61021.9.1 Predictive Maintenance Management Success Key Performance Indicators (KPIs) 61021.10 Summary 611References 61122 Analysis of PHM Patents for Electronics 613Zhenbao Liu, Zhen Jia, Chi-Man Vong, Shuhui Bu, andMichael G. Pecht22.1 Introduction 61322.2 Analysis of PHM Patents for Electronics 61622.2.1 Sources of PHM Patents 61622.2.2 Analysis of PHM Patents 61722.3 Trend of Electronics PHM 61922.3.1 Semiconductor Products and Computers 61922.3.2 Batteries 62222.3.3 Electric Motors 62622.3.4 Circuits and Systems 62922.3.5 Electrical Devices in Automobiles and Airplanes 63122.3.6 Networks and Communication Facilities 63422.3.7 Others 63622.4 Summary 638References 63923 A PHM Roadmap for Electronics-Rich Systems 64Michael G. Pecht23.1 Introduction 64923.2 Roadmap Classifications 65023.2.1 PHM at the Component Level 65123.2.1.1 PHM for Integrated Circuits 65223.2.1.2 High-Power Switching Electronics 65223.2.1.3 Built-In Prognostics for Components and Circuit Boards 65323.2.1.4 Photo-Electronics Prognostics 65423.2.1.5 Interconnect andWiring Prognostics 65623.2.2 PHM at the System Level 65723.2.2.1 Legacy Systems 65723.2.2.2 Environmental and OperationalMonitoring 65923.2.2.3 LRU to Device Level 65923.2.2.4 Dynamic Reconfiguration 65923.2.2.5 System Power Management and PHM 66023.2.2.6 PHM as Knowledge Infrastructure for System Development 66023.2.2.7 Prognostics for Software 66023.2.2.8 PHM for Mitigation of Reliability and Safety Risks 66123.2.2.9 PHM in Supply Chain Management and Product Maintenance 66223.3 Methodology Development 66323.3.1 Best Algorithms 66423.3.1.1 Approaches to Training 66723.3.1.2 Active Learning for Unlabeled Data 66723.3.1.3 Sampling Techniques and Cost-Sensitive Learning for Imbalanced Data 66823.3.1.4 Transfer Learning for Knowledge Transfer 66823.3.1.5 Internet ofThings and Big Data Analytics 66923.3.2 Verification and Validation 67023.3.3 Long-Term PHM Studies 67123.3.4 PHM for Storage 67123.3.5 PHM for No-Fault-Found/Intermittent Failures 67223.3.6 PHM for Products Subjected to Indeterminate Operating Conditions 67323.4 Nontechnical Barriers 67423.4.1 Cost, Return on Investment, and Business Case Development 67423.4.2 Liability and Litigation 67623.4.2.1 Code Architecture: Proprietary or Open? 67623.4.2.2 Long-Term Code Maintenance and Upgrades 67623.4.2.3 False Alarms, Missed Alarms, and Life-Safety Implications 67723.4.2.4 Warranty Restructuring 67723.4.3 Maintenance Culture 67723.4.4 Contract Structure 67723.4.5 Role of Standards Organizations 67823.4.5.1 IEEE Reliability Society and PHM Efforts 67823.4.5.2 SAE PHM Standards 67823.4.5.3 PHM Society 67923.4.6 Licensing and Entitlement Management 680References 680Appendix A Commercially Available Sensor Systems for PHM 691A.1 SmartButton – ACR Systems 691A.2 OWL 400 – ACR Systems 693A.3 SAVERTM 3X90 – Lansmont Instruments 695A.4 G-Link®-LXRS®– LORD MicroStrain®Sensing Systems 697A.5 V-Link®-LXRS®– LORD MicroStrain Sensing Systems 699A.6 3DM-GX4–25TM – LORD MicroStrain Sensing Systems 702A.7 IEPE-LinkTM-LXRS®– LORD MicroStrain Sensing Systems 704A.8 ICHM®20/20 – Oceana Sensor 706A.9 EnvironmentalMonitoring System 200TM – Upsite Technologies 708A.10 S2NAP®– RLWInc. 710A.11 SR1 Strain Gage Indicator – Advance Instrument Inc. 712A.12 P3 Strain Indicator and Recorder – Micro-Measurements 714A.13 Airscale Suspension-BasedWeighing System – VPG Inc. 716A.14 Radio Microlog – Transmission Dynamics 718Appendix B Journals and Conference Proceedings Related to PHM 721B.1 Journals 721B.2 Conference Proceedings 722Appendix C Glossary of Terms and Definitions 725Index 731

MICHAEL G. PECHT, PHD, is Chair Professor in Mechanical Engineering and Professor in Applied Mathematics, Statistics and Scientific Computation at the University of Maryland, USA. He is the Founder and Director of the Center for Advanced Life Cycle Engineering (CALCE) at the University of Maryland, USA, which is funded by more than 150 leading electronics companies. Dr. Pecht is an IEEE, ASME, SAE, and IMAPS Fellow and serves as editor-in-chief of IEEE Access. He has written more than 30 books, 700 technical articles, and has 8 patents. MYEONGSU KANG, PHD, is currently a Research Associate at the Center for Advanced Life Cycle Engineering (CALCE), University of Maryland, USA. His expertise is in data analytics, machine learning, system modeling, and statistics for prognostics and systems health management. He has authored/coauthored more than 60 publications in leading journals and conference proceedings.

• List of Contributors xxiiiPreface xxviiAbout the Contributors xxxvAcknowledgment xlviiList of Abbreviations xlix1 Introduction to PHM 1Michael G. Pecht andMyeongsu Kang1.1 Reliability and Prognostics 11.2 PHM for Electronics 31.3 PHM Approaches 61.3.1 PoF-Based Approach 61.3.1.1 Failure Modes, Mechanisms, and Effects Analysis (FMMEA) 71.3.1.2 Life-Cycle Load Monitoring 81.3.1.3 Data Reduction and Load Feature Extraction 101.3.1.4 Data Assessment and Remaining Life Calculation 121.3.1.5 Uncertainty Implementation and Assessment 131.3.2 Canaries 141.3.3 Data-Driven Approach 161.3.3.1 Monitoring and Reasoning of Failure Precursors 161.3.3.2 Data Analytics and Machine Learning 201.3.4 Fusion Approach 231.4 Implementation of PHM in a System of Systems 241.5 PHM in the Internet ofThings (IoT) Era 261.5.1 IoT-Enabled PHM Applications: Manufacturing 271.5.2 IoT-Enabled PHM Applications: Energy Generation 271.5.3 IoT-Enabled PHM Applications: Transportation and Logistics 281.5.4 IoT-Enabled PHM Applications: Automobiles 281.5.5 IoT-Enabled PHM Applications: Medical Consumer Products 291.5.6 IoT-Enabled PHM Applications:Warranty Services 291.5.7 IoT-Enabled PHM Applications: Robotics 301.6 Summary 30References 302 Sensor Systems for PHM 39Hyunseok Oh,Michael H. Azarian, Shunfeng Cheng, andMichael G. Pecht2.1 Sensor and Sensing Principles 392.1.1 Thermal Sensors 402.1.2 Electrical Sensors 412.1.3 Mechanical Sensors 422.1.4 Chemical Sensors 422.1.5 Humidity Sensors 442.1.6 Biosensors 442.1.7 Optical Sensors 452.1.8 Magnetic Sensors 452.2 Sensor Systems for PHM 462.2.1 Parameters to be Monitored 472.2.2 Sensor System Performance 482.2.3 Physical Attributes of Sensor Systems 482.2.4 Functional Attributes of Sensor Systems 492.2.4.1 Onboard Power and Power Management 492.2.4.2 Onboard Memory and Memory Management 502.2.4.3 Programmable SamplingMode and Sampling Rate 512.2.4.4 Signal Processing Software 512.2.4.5 Fast and Convenient Data Transmission 522.2.5 Reliability 532.2.6 Availability 532.2.7 Cost 542.3 Sensor Selection 542.4 Examples of Sensor Systems for PHM Implementation 542.5 Emerging Trends in Sensor Technology for PHM 59References 603 Physics-of-Failure Approach to PHM 61Shunfeng Cheng, Nagarajan Raghavan, Jie Gu, Sony Mathew, and Michael G. Pecht3.1 PoF-Based PHM Methodology 613.2 Hardware Configuration 623.3 Loads 633.4 Failure Modes, Mechanisms, and Effects Analysis (FMMEA) 643.4.1 Examples of FMMEA for Electronic Devices 683.5 Stress Analysis 713.6 Reliability Assessment and Remaining-Life Predictions 733.7 Outputs from PoF-Based PHM 773.8 Caution and Concerns in the Use of PoF-Based PHM 783.9 Combining PoF with Data-Driven Prognosis 80References 814 Machine Learning: Fundamentals 85Myeongsu Kang and Noel Jordan Jameson4.1 Types of Machine Learning 854.1.1 Supervised, Unsupervised, Semi-Supervised, and Reinforcement Learning 864.1.2 Batch and Online Learning 884.1.3 Instance-Based and Model-Based Learning 894.2 Probability Theory in Machine Learning: Fundamentals 904.2.1 Probability Space and Random Variables 914.2.2 Distributions, Joint Distributions, and Marginal Distributions 914.2.3 Conditional Distributions 914.2.4 Independence 924.2.5 Chain Rule and Bayes Rule 924.3 Probability Mass Function and Probability Density Function 934.3.1 Probability Mass Function 934.3.2 Probability Density Function 934.4 Mean, Variance, and Covariance Estimation 944.4.1 Mean 944.4.2 Variance 944.4.3 Robust Covariance Estimation 954.5 Probability Distributions 964.5.1 Bernoulli Distribution 964.5.2 Normal Distribution 964.5.3 Uniform Distribution 974.6 Maximum Likelihood and Maximum A Posteriori Estimation 974.6.1 Maximum Likelihood Estimation 974.6.2 Maximum A Posteriori Estimation 984.7 Correlation and Causation 994.8 Kernel Trick 1004.9 Performance Metrics 1024.9.1 Diagnostic Metrics 1024.9.2 Prognostic Metrics 105References 1075 Machine Learning: Data Pre-processing 111Myeongsu Kang and Jing Tian5.1 Data Cleaning 1115.1.1 Missing Data Handling 1115.1.1.1 Single-Value Imputation Methods 1135.1.1.2 Model-Based Methods 1135.2 Feature Scaling 1145.3 Feature Engineering 1165.3.1 Feature Extraction 1165.3.1.1 PCA and Kernel PCA 1165.3.1.2 LDA and Kernel LDA 1185.3.1.3 Isomap 1195.3.1.4 Self-Organizing Map (SOM) 1205.3.2 Feature Selection 1215.3.2.1 Feature Selection: FilterMethods 1225.3.2.2 Feature Selection:WrapperMethods 1245.3.2.3 Feature Selection: Embedded Methods 1245.3.2.4 Advanced Feature Selection 1255.4 Imbalanced Data Handling 1255.4.1 SamplingMethods for Imbalanced Learning 1265.4.1.1 Synthetic Minority Oversampling Technique 1265.4.1.2 Adaptive Synthetic Sampling 1265.4.1.3 Effect of SamplingMethods for Diagnosis 127References 1296 Machine Learning: Anomaly Detection 131Myeongsu Kang6.1 Introduction 1316.2 Types of Anomalies 1336.2.1 Point Anomalies 1346.2.2 Contextual Anomalies 1346.2.3 Collective Anomalies 1356.3 Distance-Based Methods 1366.3.1 MD Calculation Using an Inverse Matrix Method 1376.3.2 MD Calculation Using a Gram–Schmidt Orthogonalization Method 1376.3.3 Decision Rules 1386.3.3.1 Gamma Distribution:Threshold Selection 1386.3.3.2 Weibull Distribution:Threshold Selection 1396.3.3.3 Box-Cox Transformation:Threshold Selection 1396.4 Clustering-Based Methods 1406.4.1 k-Means Clustering 1416.4.2 Fuzzy c-Means Clustering 1426.4.3 Self-Organizing Maps (SOMs) 1426.5 Classification-Based Methods 1446.5.1 One-Class Classification 1456.5.1.1 One-Class Support Vector Machines 1456.5.1.2 k-Nearest Neighbors 1486.5.2 Multi-Class Classification 1496.5.2.1 Multi-Class Support Vector Machines 1496.5.2.2 Neural Networks 1516.6 StatisticalMethods 1536.6.1 Sequential Probability Ratio Test 1546.6.2 Correlation Analysis 1566.7 Anomaly Detection with No System Health Profile 1566.8 Challenges in Anomaly Detection 158References 1597 Machine Learning: Diagnostics and Prognostics 163Myeongsu Kang7.1 Overview of Diagnosis and Prognosis 1637.2 Techniques for Diagnostics 1657.2.1 Supervised Machine Learning Algorithms 1657.2.1.1 Naïve Bayes 1657.2.1.2 Decision Trees 1677.2.2 Ensemble Learning 1697.2.2.1 Bagging 1707.2.2.2 Boosting: AdaBoost 1717.2.3 Deep Learning 1727.2.3.1 Supervised Learning: Deep Residual Networks 1737.2.3.2 Effect of Feature Learning-Powered Diagnosis 1767.3 Techniques for Prognostics 1787.3.1 Regression Analysis 1787.3.1.1 Linear Regression 1787.3.1.2 Polynomial Regression 1807.3.1.3 Ridge Regression 1817.3.1.4 LASSO Regression 1827.3.1.5 Elastic Net Regression 1837.3.1.6 k-Nearest Neighbors Regression 1837.3.1.7 Support Vector Regression 1847.3.2 Particle Filtering 1857.3.2.1 Fundamentals of Particle Filtering 1867.3.2.2 Resampling Methods – A Review 187References 1898 Uncertainty Representation, Quantification, and Management in Prognostics 193Shankar Sankararaman8.1 Introduction 1938.2 Sources of Uncertainty in PHM 1968.3 Formal Treatment of Uncertainty in PHM 1998.3.1 Problem 1: Uncertainty Representation and Interpretation 1998.3.2 Problem 2: Uncertainty Quantification 1998.3.3 Problem 3: Uncertainty Propagation 2008.3.4 Problem 4: Uncertainty Management 2008.4 Uncertainty Representation and Interpretation 2008.4.1 Physical Probabilities and Testing-Based Prediction 2018.4.1.1 Physical Probability 2018.4.1.2 Testing-Based Life Prediction 2018.4.1.3 Confidence Intervals 2028.4.2 Subjective Probabilities and Condition-Based Prognostics 2028.4.2.1 Subjective Probability 2028.4.2.2 Subjective Probabilities in Condition-Based Prognostics 2038.4.3 Why is RUL Prediction Uncertain? 2038.5 Uncertainty Quantification and Propagation for RUL Prediction 2038.5.1 Computational Framework for Uncertainty Quantification 2048.5.1.1 Present State Estimation 2048.5.1.2 Future State Prediction 2058.5.1.3 RUL Computation 2058.5.2 RUL Prediction: An Uncertainty Propagation Problem 2068.5.3 Uncertainty PropagationMethods 2068.5.3.1 Sampling-Based Methods 2078.5.3.2 AnalyticalMethods 2098.5.3.3 Hybrid Methods 2098.5.3.4 Summary of Methods 2098.6 Uncertainty Management 2108.7 Case Study: Uncertainty Quantification in the Power System of an Unmanned Aerial Vehicle 2118.7.1 Description of the Model 2118.7.2 Sources of Uncertainty 2128.7.3 Results: Constant Amplitude Loading Conditions 2138.7.4 Results: Variable Amplitude Loading Conditions 2148.7.5 Discussion 2148.8 Existing Challenges 2158.8.1 Timely Predictions 2158.8.2 Uncertainty Characterization 2168.8.3 Uncertainty Propagation 2168.8.4 Capturing Distribution Properties 2168.8.5 Accuracy 2168.8.6 Uncertainty Bounds 2168.8.7 Deterministic Calculations 2168.9 Summary 217References 2179 PHM Cost and Return on Investment 221Peter Sandborn, ChrisWilkinson, Kiri Lee Sharon, Taoufik Jazouli, and Roozbeh Bakhshi9.1 Return on Investment 2219.1.1 PHM ROI Analyses 2229.1.2 Financial Costs 2249.2 PHM Cost-Modeling Terminology and Definitions 2259.3 PHM Implementation Costs 2269.3.1 Nonrecurring Costs 2269.3.2 Recurring Costs 2279.3.3 Infrastructure Costs 2289.3.4 Nonmonetary Considerations and Maintenance Culture 2289.4 Cost Avoidance 2299.4.1 Maintenance Planning Cost Avoidance 2319.4.2 Discrete-Event Simulation Maintenance PlanningModel 2329.4.3 Fixed-Schedule Maintenance Interval 2339.4.4 Data-Driven (Precursor to Failure Monitoring) Methods 2339.4.5 Model-Based (LRU-Independent)Methods 2349.4.6 Discrete-Event Simulation Implementation Details 2369.4.7 Operational Profile 2379.5 Example PHM Cost Analysis 2389.5.1 Single-Socket Model Results 2399.5.2 Multiple-Socket Model Results 2419.6 Example Business Case Construction: Analysis for ROI 2469.7 Summary 255References 25510 Valuation and Optimization of PHM-Enabled Maintenance Decisions 261Xin Lei, Amir Reza Kashani-Pour, Peter Sandborn, and Taoufik Jazouli10.1 Valuation and Optimization of PHM-Enabled Maintenance Decisions for an Individual System 26210.1.1 A PHM-Enabled Predictive Maintenance OptimizationModel for an Individual System 26310.1.2 Case Study: Optimization of PHM-Enabled Maintenance Decisions for an Individual System (Wind Turbine) 26510.2 Availability 26810.2.1 The Business of Availability: Outcome-Based Contracts 26910.2.2 Incorporating Contract Terms into Maintenance Decisions 27010.2.3 Case Study: Optimization of PHM-Enabled Maintenance Decisions for Systems (Wind Farm) 27010.3 Future Directions 27210.3.1 Design for Availability 27210.3.2 Prognostics-BasedWarranties 27510.3.3 Contract Engineering 276References 27711 Health and Remaining Useful Life Estimation of Electronic Circuits 279Arvind Sai Sarathi Vasan and Michael G. Pecht11.1 Introduction 27911.2 RelatedWork 28111.2.1 Component-Centric Approach 28111.2.2 Circuit-Centric Approach 28211.3 Electronic Circuit Health Estimation Through Kernel Learning 28511.3.1 Kernel-Based Learning 28511.3.2 Health Estimation Method 28611.3.2.1 Likelihood-Based Function for Model Selection 28811.3.2.2 Optimization Approach for Model Selection 28911.3.3 Implementation Results 29211.3.3.1 Bandpass Filter Circuit 29311.3.3.2 DC–DC Buck Converter System 30011.4 RUL Prediction Using Model-Based Filtering 30611.4.1 Prognostics Problem Formulation 30611.4.2 Circuit DegradationModeling 30711.4.3 Model-Based Prognostic Methodology 31011.4.4 Implementation Results 31311.4.4.1 Low-Pass Filter Circuit 31311.4.4.2 Voltage Feedback Circuit 31511.4.4.3 Source of RUL Prediction Error 32011.4.4.4 Effect of First-Principles-Based Modeling 32011.5 Summary 322References 32412 PHM-Based Qualification of Electronics 329Preeti S. Chauhan12.1 Why is Product Qualification Important? 32912.2 Considerations for Product Qualification 33112.3 Review of Current Qualification Methodologies 33412.3.1 Standards-Based Qualification 33412.3.2 Knowledge-Based or PoF-Based Qualification 33712.3.3 Prognostics and Health Management-Based Qualification 34012.3.3.1 Data-Driven Techniques 34012.3.3.2 Fusion Prognostics 34312.4 Summary 345References 34613 PHM of Li-ion Batteries 349Saurabh Saxena, Yinjiao Xing, andMichael G. Pecht13.1 Introduction 34913.2 State of Charge Estimation 35113.2.1 SOC Estimation Case Study I 35213.2.1.1 NN Model 35313.2.1.2 Training and Testing Data 35413.2.1.3 Determination of the NN Structure 35513.2.1.4 Training and Testing Results 35613.2.1.5 Application of Unscented Kalman Filter 35713.2.2 SOC Estimation Case Study II 35713.2.2.1 OCV–SOC-T Test 35813.2.2.2 Battery Modeling and Parameter Identification 35913.2.2.3 OCV–SOC-T Table for Model Improvement 36013.2.2.4 Validation of the Proposed Model 36213.2.2.5 Algorithm Implementation for Online Estimation 36213.3 State of Health Estimation and Prognostics 36513.3.1 Case Study for Li-ion Battery Prognostics 36613.3.1.1 Capacity DegradationModel 36613.3.1.2 Uncertainties in Battery Prognostics 36813.3.1.3 Model Updating via Bayesian Monte Carlo 36813.3.1.4 SOH Prognostics and RUL Estimation 36913.3.1.5 Prognostic Results 37113.4 Summary 371References 37214 PHM of Light-Emitting Diodes 377Moon-Hwan Chang, Jiajie Fan, Cheng Qian, and Bo Sun14.1 Introduction 37714.2 Review of PHM Methodologies for LEDs 37814.2.1 Overview of Available Prognostic Methods 37814.2.2 Data-DrivenMethods 37914.2.2.1 Statistical Regression 37914.2.2.2 Static Bayesian Network 38114.2.2.3 Kalman Filtering 38214.2.2.4 Particle Filtering 38314.2.2.5 Artificial Neural Network 38414.2.3 Physics-Based Methods 38514.2.4 LED System-Level Prognostics 38714.3 Simulation-Based Modeling and Failure Analysis for LEDs 38814.3.1 LED Chip-LevelModeling and Failure Analysis 38914.3.1.1 Electro-optical Simulation of LED Chip 38914.3.1.2 LED Chip-Level Failure Analysis 39314.3.2 LED Package-Level Modeling and Failure Analysis 39514.3.2.1 Thermal and Optical Simulation for Phosphor-Converted White LED Package 39514.3.2.2 LED Package-Level Failure Analysis 39714.3.3 LED System-LevelModeling and Failure Analysis 39914.4 Return-on-Investment Analysis of Applying Health Monitoring to LED Lighting Systems 40114.4.1 ROI Methodology 40314.4.2 ROI Analysis of Applying System Health Monitoring to LED Lighting Systems 40614.4.2.1 Failure Rates and Distributions for ROI Simulation 40714.4.2.2 Determination of Prognostics Distance 41014.4.2.3 IPHM, CPHM, and Cu Evaluation 41214.4.2.4 ROI Evaluation 41714.5 Summary 419References 42015 PHM in Healthcare 431Mary Capelli-Schellpfeffer,Myeongsu Kang, andMichael G. Pecht15.1 Healthcare in the United States 43115.2 Considerations in Healthcare 43215.2.1 Clinical Consideration in ImplantableMedical Devices 43215.2.2 Considerations in Care Bots 43315.3 Benefits of PHM 43815.3.1 Safety Increase 43915.3.2 Operational Reliability Improvement 44015.3.3 Mission Availability Increase 44015.3.4 System’s Service Life Extension 44115.3.5 Maintenance Effectiveness Increase 44115.4 PHM of ImplantableMedical Devices 44215.5 PHM of Care Bots 44415.6 Canary-Based Prognostics of Healthcare Devices 44515.7 Summary 447References 44716 PHM of Subsea Cables 451David Flynn, Christopher Bailey, Pushpa Rajaguru,Wenshuo Tang, and Chunyan Yin16.1 Subsea Cable Market 45116.2 Subsea Cables 45216.3 Cable Failures 45416.3.1 Internal Failures 45516.3.2 Early-Stage Failures 45516.3.3 External Failures 45516.3.4 Environmental Conditions 45516.3.5 Third-Party Damage 45616.4 State-of-the-Art Monitoring 45716.5 Qualifying and Maintaining Subsea Cables 45816.5.1 Qualifying Subsea Cables 45816.5.2 Mechanical Tests 45816.5.3 Maintaining Subsea Cables 45916.6 Data-Gathering Techniques 46016.7 Measuring theWear Behavior of Cable Materials 46116.8 Predicting Cable Movement 46316.8.1 Sliding Distance Derivation 46316.8.2 Scouring Depth Calculations 46516.9 Predicting Cable Degradation 46616.9.1 Volume Loss due to Abrasion 46616.9.2 Volume Loss due to Corrosion 46616.10 Predicting Remaining Useful Life 46816.11 Case Study 47116.12 Future Challenges 47116.12.1 Data-Driven Approach for Random Failures 47116.12.2 Model-Driven Approach for Environmental Failures 47316.12.2.1 Fusion-Based PHM 47316.12.2.2 Sensing Techniques 47416.13 Summary 474References 47517 Connected Vehicle Diagnostics and Prognostics 479Yilu Zhang and Xinyu Du17.1 Introduction 47917.2 Design of an Automatic Field Data Analyzer 48117.2.1 Data Collection Subsystem 48217.2.2 Information Abstraction Subsystem 48217.2.3 Root Cause Analysis Subsystem 48217.2.3.1 Feature-Ranking Module 48217.2.3.2 Relevant Feature Set Selection 48417.2.3.3 Results Interpretation 48617.3 Case Study: CVDP for Vehicle Batteries 48617.3.1 Brief Background of Vehicle Batteries 48617.3.2 Applying AFDA for Vehicle Batteries 48817.3.3 Experimental Results 489Contents xvii17.3.3.1 Information Abstraction 49017.3.3.2 Feature Ranking 49017.3.3.3 Interpretation of Results 49517.4 Summary 498References 49918 The Role of PHM at Commercial Airlines 503RhondaWalthall and Ravi Rajamani18.1 Evolution of Aviation Maintenance 50318.2 Stakeholder Expectations for PHM 50618.2.1 Passenger Expectations 50618.2.2 Airline/Operator/Owner Expectations 50718.2.3 Airframe Manufacturer Expectations 50918.2.4 Engine Manufacturer Expectations 51018.2.5 System and Component Supplier Expectations 51118.2.6 MRO Organization Expectations 51218.3 PHM Implementation 51318.3.1 SATAA 51318.4 PHM Applications 51718.4.1 Engine Health Management (EHM) 51718.4.1.1 History of EHM 51818.4.1.2 EHM Infrastructure 51918.4.1.3 Technologies Associated with EHM 52018.4.1.4 The Future 52318.4.2 Auxiliary Power Unit (APU) Health Management 52418.4.3 Environmental Control System (ECS) and Air Distribution Health Monitoring 52518.4.4 Landing System Health Monitoring 52618.4.5 Liquid Cooling System Health Monitoring 52618.4.6 Nitrogen Generation System (NGS) Health Monitoring 52718.4.7 Fuel Consumption Monitoring 52718.4.8 Flight Control Actuation Health Monitoring 52818.4.9 Electric Power System Health Monitoring 52918.4.10 Structural Health Monitoring (SHM) 52918.4.11 Battery Health Management 53118.5 Summary 532References 53319 PHM Software for Electronics 535Noel Jordan Jameson,Myeongsu Kang, and Jing Tian19.1 PHM Software: CALCE Simulation Assisted Reliability Assessment 53519.2 PHM Software: Data-Driven 54019.2.1 Data Flow 54119.2.2 Master Options 54219.2.3 Data Pre-processing 54319.2.4 Feature Discovery 54519.2.5 Anomaly Detection 54619.2.6 Diagnostics/Classification 54819.2.7 Prognostics/Modeling 55219.2.8 Challenges in Data-Driven PHM Software Development 55419.3 Summary 55720 eMaintenance 559Ramin Karim, Phillip Tretten, and Uday Kumar20.1 From Reactive to Proactive Maintenance 55920.2 The Onset of eMaintenance 56020.3 MaintenanceManagement System 56120.3.1 Life-cycle Management 56220.3.2 eMaintenance Architecture 56420.4 Sensor Systems 56420.4.1 Sensor Technology for PHM 56520.5 Data Analysis 56520.6 Predictive Maintenance 56620.7 Maintenance Analytics 56720.7.1 Maintenance Descriptive Analytics 56820.7.2 Maintenance Analytics and eMaintenance 56820.7.3 Maintenance Analytics and Big Data 56820.8 Knowledge Discovery 57020.9 Integrated Knowledge Discovery 57120.10 User Interface for Decision Support 57220.11 Applications of eMaintenance 57220.11.1 eMaintenance in Railways 57220.11.1.1 Railway Cloud: Swedish Railway Data 57320.11.1.2 Railway Cloud: Service Architecture 57320.11.1.3 Railway Cloud: Usage Scenario 57420.11.2 eMaintenance in Manufacturing 57420.11.3 MEMS Sensors for Bearing Vibration Measurement 57620.11.4 Wireless Sensors for Temperature Measurement 57620.11.5 Monitoring Systems 57620.11.6 eMaintenance Cloud and Servers 57820.11.7 Dashboard Managers 58020.11.8 Alarm Servers 58020.11.9 Cloud Services 58120.11.10 Graphic User Interfaces 58320.12 Internet Technology and Optimizing Technology 585References 58621 Predictive Maintenance in the IoT Era 589Rashmi B. Shetty21.1 Background 58921.1.1 Challenges of a Maintenance Program 59021.1.2 Evolution of Maintenance Paradigms 59021.1.3 Preventive Versus Predictive Maintenance 59221.1.4 P–F Curve 59221.1.5 Bathtub Curve 59421.2 Benefits of a Predictive Maintenance Program 59521.3 Prognostic Model Selection for Predictive Maintenance 59621.4 Internet ofThings 59821.4.1 Industrial IoT 59821.5 Predictive Maintenance Based on IoT 59921.6 Predictive Maintenance Usage Cases 60021.7 Machine Learning Techniques for Data-Driven Predictive Maintenance 60021.7.1 Supervised Learning 60221.7.2 Unsupervised Learning 60221.7.3 Anomaly Detection 60221.7.4 Multi-class and Binary Classification Models 60321.7.5 Regression Models 60421.7.6 Survival Models 60421.8 Best Practices 60421.8.1 Define Business Problem and QuantitativeMetrics 60521.8.2 Identify Assets and Data Sources 60521.8.3 Data Acquisition and Transformation 60621.8.4 Build Models 60721.8.5 Model Selection 60721.8.6 Predict Outcomes and Transform into Process Insights 60821.8.7 Operationalize and Deploy 60921.8.8 Continuous Monitoring 60921.9 Challenges in a Successful Predictive Maintenance Program 61021.9.1 Predictive Maintenance Management Success Key Performance Indicators (KPIs) 61021.10 Summary 611References 61122 Analysis of PHM Patents for Electronics 613Zhenbao Liu, Zhen Jia, Chi-Man Vong, Shuhui Bu, andMichael G. Pecht22.1 Introduction 61322.2 Analysis of PHM Patents for Electronics 61622.2.1 Sources of PHM Patents 61622.2.2 Analysis of PHM Patents 61722.3 Trend of Electronics PHM 61922.3.1 Semiconductor Products and Computers 61922.3.2 Batteries 62222.3.3 Electric Motors 62622.3.4 Circuits and Systems 62922.3.5 Electrical Devices in Automobiles and Airplanes 63122.3.6 Networks and Communication Facilities 63422.3.7 Others 63622.4 Summary 638References 63923 A PHM Roadmap for Electronics-Rich Systems 64Michael G. Pecht23.1 Introduction 64923.2 Roadmap Classifications 65023.2.1 PHM at the Component Level 65123.2.1.1 PHM for Integrated Circuits 65223.2.1.2 High-Power Switching Electronics 65223.2.1.3 Built-In Prognostics for Components and Circuit Boards 65323.2.1.4 Photo-Electronics Prognostics 65423.2.1.5 Interconnect andWiring Prognostics 65623.2.2 PHM at the System Level 65723.2.2.1 Legacy Systems 65723.2.2.2 Environmental and OperationalMonitoring 65923.2.2.3 LRU to Device Level 65923.2.2.4 Dynamic Reconfiguration 65923.2.2.5 System Power Management and PHM 66023.2.2.6 PHM as Knowledge Infrastructure for System Development 66023.2.2.7 Prognostics for Software 66023.2.2.8 PHM for Mitigation of Reliability and Safety Risks 66123.2.2.9 PHM in Supply Chain Management and Product Maintenance 66223.3 Methodology Development 66323.3.1 Best Algorithms 66423.3.1.1 Approaches to Training 66723.3.1.2 Active Learning for Unlabeled Data 66723.3.1.3 Sampling Techniques and Cost-Sensitive Learning for Imbalanced Data 66823.3.1.4 Transfer Learning for Knowledge Transfer 66823.3.1.5 Internet ofThings and Big Data Analytics 66923.3.2 Verification and Validation 67023.3.3 Long-Term PHM Studies 67123.3.4 PHM for Storage 67123.3.5 PHM for No-Fault-Found/Intermittent Failures 67223.3.6 PHM for Products Subjected to Indeterminate Operating Conditions 67323.4 Nontechnical Barriers 67423.4.1 Cost, Return on Investment, and Business Case Development 67423.4.2 Liability and Litigation 67623.4.2.1 Code Architecture: Proprietary or Open? 67623.4.2.2 Long-Term Code Maintenance and Upgrades 67623.4.2.3 False Alarms, Missed Alarms, and Life-Safety Implications 67723.4.2.4 Warranty Restructuring 67723.4.3 Maintenance Culture 67723.4.4 Contract Structure 67723.4.5 Role of Standards Organizations 67823.4.5.1 IEEE Reliability Society and PHM Efforts 67823.4.5.2 SAE PHM Standards 67823.4.5.3 PHM Society 67923.4.6 Licensing and Entitlement Management 680References 680Appendix A Commercially Available Sensor Systems for PHM 691A.1 SmartButton – ACR Systems 691A.2 OWL 400 – ACR Systems 693A.3 SAVERTM 3X90 – Lansmont Instruments 695A.4 G-Link®-LXRS®– LORD MicroStrain®Sensing Systems 697A.5 V-Link®-LXRS®– LORD MicroStrain Sensing Systems 699A.6 3DM-GX4–25TM – LORD MicroStrain Sensing Systems 702A.7 IEPE-LinkTM-LXRS®– LORD MicroStrain Sensing Systems 704A.8 ICHM®20/20 – Oceana Sensor 706A.9 EnvironmentalMonitoring System 200TM – Upsite Technologies 708A.10 S2NAP®– RLWInc. 710A.11 SR1 Strain Gage Indicator – Advance Instrument Inc. 712A.12 P3 Strain Indicator and Recorder – Micro-Measurements 714A.13 Airscale Suspension-BasedWeighing System – VPG Inc. 716A.14 Radio Microlog – Transmission Dynamics 718Appendix B Journals and Conference Proceedings Related to PHM 721B.1 Journals 721B.2 Conference Proceedings 722Appendix C Glossary of Terms and Definitions 725Index 731

MICHAEL G. PECHT, PHD, is Chair Professor in Mechanical Engineering and Professor in Applied Mathematics, Statistics and Scientific Computation at the University of Maryland, USA. He is the Founder and Director of the Center for Advanced Life Cycle Engineering (CALCE) at the University of Maryland, USA, which is funded by more than 150 leading electronics companies. Dr. Pecht is an IEEE, ASME, SAE, and IMAPS Fellow and serves as editor-in-chief of IEEE Access. He has written more than 30 books, 700 technical articles, and has 8 patents. MYEONGSU KANG, PHD, is currently a Research Associate at the Center for Advanced Life Cycle Engineering (CALCE), University of Maryland, USA. His expertise is in data analytics, machine learning, system modeling, and statistics for prognostics and systems health management. He has authored/coauthored more than 60 publications in leading journals and conference proceedings.

• List of Contributors xxiiiPreface xxviiAbout the Contributors xxxvAcknowledgment xlviiList of Abbreviations xlix1 Introduction to PHM 1Michael G. Pecht andMyeongsu Kang1.1 Reliability and Prognostics 11.2 PHM for Electronics 31.3 PHM Approaches 61.3.1 PoF-Based Approach 61.3.1.1 Failure Modes, Mechanisms, and Effects Analysis (FMMEA) 71.3.1.2 Life-Cycle Load Monitoring 81.3.1.3 Data Reduction and Load Feature Extraction 101.3.1.4 Data Assessment and Remaining Life Calculation 121.3.1.5 Uncertainty Implementation and Assessment 131.3.2 Canaries 141.3.3 Data-Driven Approach 161.3.3.1 Monitoring and Reasoning of Failure Precursors 161.3.3.2 Data Analytics and Machine Learning 201.3.4 Fusion Approach 231.4 Implementation of PHM in a System of Systems 241.5 PHM in the Internet ofThings (IoT) Era 261.5.1 IoT-Enabled PHM Applications: Manufacturing 271.5.2 IoT-Enabled PHM Applications: Energy Generation 271.5.3 IoT-Enabled PHM Applications: Transportation and Logistics 281.5.4 IoT-Enabled PHM Applications: Automobiles 281.5.5 IoT-Enabled PHM Applications: Medical Consumer Products 291.5.6 IoT-Enabled PHM Applications:Warranty Services 291.5.7 IoT-Enabled PHM Applications: Robotics 301.6 Summary 30References 302 Sensor Systems for PHM 39Hyunseok Oh,Michael H. Azarian, Shunfeng Cheng, andMichael G. Pecht2.1 Sensor and Sensing Principles 392.1.1 Thermal Sensors 402.1.2 Electrical Sensors 412.1.3 Mechanical Sensors 422.1.4 Chemical Sensors 422.1.5 Humidity Sensors 442.1.6 Biosensors 442.1.7 Optical Sensors 452.1.8 Magnetic Sensors 452.2 Sensor Systems for PHM 462.2.1 Parameters to be Monitored 472.2.2 Sensor System Performance 482.2.3 Physical Attributes of Sensor Systems 482.2.4 Functional Attributes of Sensor Systems 492.2.4.1 Onboard Power and Power Management 492.2.4.2 Onboard Memory and Memory Management 502.2.4.3 Programmable SamplingMode and Sampling Rate 512.2.4.4 Signal Processing Software 512.2.4.5 Fast and Convenient Data Transmission 522.2.5 Reliability 532.2.6 Availability 532.2.7 Cost 542.3 Sensor Selection 542.4 Examples of Sensor Systems for PHM Implementation 542.5 Emerging Trends in Sensor Technology for PHM 59References 603 Physics-of-Failure Approach to PHM 61Shunfeng Cheng, Nagarajan Raghavan, Jie Gu, Sony Mathew, and Michael G. Pecht3.1 PoF-Based PHM Methodology 613.2 Hardware Configuration 623.3 Loads 633.4 Failure Modes, Mechanisms, and Effects Analysis (FMMEA) 643.4.1 Examples of FMMEA for Electronic Devices 683.5 Stress Analysis 713.6 Reliability Assessment and Remaining-Life Predictions 733.7 Outputs from PoF-Based PHM 773.8 Caution and Concerns in the Use of PoF-Based PHM 783.9 Combining PoF with Data-Driven Prognosis 80References 814 Machine Learning: Fundamentals 85Myeongsu Kang and Noel Jordan Jameson4.1 Types of Machine Learning 854.1.1 Supervised, Unsupervised, Semi-Supervised, and Reinforcement Learning 864.1.2 Batch and Online Learning 884.1.3 Instance-Based and Model-Based Learning 894.2 Probability Theory in Machine Learning: Fundamentals 904.2.1 Probability Space and Random Variables 914.2.2 Distributions, Joint Distributions, and Marginal Distributions 914.2.3 Conditional Distributions 914.2.4 Independence 924.2.5 Chain Rule and Bayes Rule 924.3 Probability Mass Function and Probability Density Function 934.3.1 Probability Mass Function 934.3.2 Probability Density Function 934.4 Mean, Variance, and Covariance Estimation 944.4.1 Mean 944.4.2 Variance 944.4.3 Robust Covariance Estimation 954.5 Probability Distributions 964.5.1 Bernoulli Distribution 964.5.2 Normal Distribution 964.5.3 Uniform Distribution 974.6 Maximum Likelihood and Maximum A Posteriori Estimation 974.6.1 Maximum Likelihood Estimation 974.6.2 Maximum A Posteriori Estimation 984.7 Correlation and Causation 994.8 Kernel Trick 1004.9 Performance Metrics 1024.9.1 Diagnostic Metrics 1024.9.2 Prognostic Metrics 105References 1075 Machine Learning: Data Pre-processing 111Myeongsu Kang and Jing Tian5.1 Data Cleaning 1115.1.1 Missing Data Handling 1115.1.1.1 Single-Value Imputation Methods 1135.1.1.2 Model-Based Methods 1135.2 Feature Scaling 1145.3 Feature Engineering 1165.3.1 Feature Extraction 1165.3.1.1 PCA and Kernel PCA 1165.3.1.2 LDA and Kernel LDA 1185.3.1.3 Isomap 1195.3.1.4 Self-Organizing Map (SOM) 1205.3.2 Feature Selection 1215.3.2.1 Feature Selection: FilterMethods 1225.3.2.2 Feature Selection:WrapperMethods 1245.3.2.3 Feature Selection: Embedded Methods 1245.3.2.4 Advanced Feature Selection 1255.4 Imbalanced Data Handling 1255.4.1 SamplingMethods for Imbalanced Learning 1265.4.1.1 Synthetic Minority Oversampling Technique 1265.4.1.2 Adaptive Synthetic Sampling 1265.4.1.3 Effect of SamplingMethods for Diagnosis 127References 1296 Machine Learning: Anomaly Detection 131Myeongsu Kang6.1 Introduction 1316.2 Types of Anomalies 1336.2.1 Point Anomalies 1346.2.2 Contextual Anomalies 1346.2.3 Collective Anomalies 1356.3 Distance-Based Methods 1366.3.1 MD Calculation Using an Inverse Matrix Method 1376.3.2 MD Calculation Using a Gram–Schmidt Orthogonalization Method 1376.3.3 Decision Rules 1386.3.3.1 Gamma Distribution:Threshold Selection 1386.3.3.2 Weibull Distribution:Threshold Selection 1396.3.3.3 Box-Cox Transformation:Threshold Selection 1396.4 Clustering-Based Methods 1406.4.1 k-Means Clustering 1416.4.2 Fuzzy c-Means Clustering 1426.4.3 Self-Organizing Maps (SOMs) 1426.5 Classification-Based Methods 1446.5.1 One-Class Classification 1456.5.1.1 One-Class Support Vector Machines 1456.5.1.2 k-Nearest Neighbors 1486.5.2 Multi-Class Classification 1496.5.2.1 Multi-Class Support Vector Machines 1496.5.2.2 Neural Networks 1516.6 StatisticalMethods 1536.6.1 Sequential Probability Ratio Test 1546.6.2 Correlation Analysis 1566.7 Anomaly Detection with No System Health Profile 1566.8 Challenges in Anomaly Detection 158References 1597 Machine Learning: Diagnostics and Prognostics 163Myeongsu Kang7.1 Overview of Diagnosis and Prognosis 1637.2 Techniques for Diagnostics 1657.2.1 Supervised Machine Learning Algorithms 1657.2.1.1 Naïve Bayes 1657.2.1.2 Decision Trees 1677.2.2 Ensemble Learning 1697.2.2.1 Bagging 1707.2.2.2 Boosting: AdaBoost 1717.2.3 Deep Learning 1727.2.3.1 Supervised Learning: Deep Residual Networks 1737.2.3.2 Effect of Feature Learning-Powered Diagnosis 1767.3 Techniques for Prognostics 1787.3.1 Regression Analysis 1787.3.1.1 Linear Regression 1787.3.1.2 Polynomial Regression 1807.3.1.3 Ridge Regression 1817.3.1.4 LASSO Regression 1827.3.1.5 Elastic Net Regression 1837.3.1.6 k-Nearest Neighbors Regression 1837.3.1.7 Support Vector Regression 1847.3.2 Particle Filtering 1857.3.2.1 Fundamentals of Particle Filtering 1867.3.2.2 Resampling Methods – A Review 187References 1898 Uncertainty Representation, Quantification, and Management in Prognostics 193Shankar Sankararaman8.1 Introduction 1938.2 Sources of Uncertainty in PHM 1968.3 Formal Treatment of Uncertainty in PHM 1998.3.1 Problem 1: Uncertainty Representation and Interpretation 1998.3.2 Problem 2: Uncertainty Quantification 1998.3.3 Problem 3: Uncertainty Propagation 2008.3.4 Problem 4: Uncertainty Management 2008.4 Uncertainty Representation and Interpretation 2008.4.1 Physical Probabilities and Testing-Based Prediction 2018.4.1.1 Physical Probability 2018.4.1.2 Testing-Based Life Prediction 2018.4.1.3 Confidence Intervals 2028.4.2 Subjective Probabilities and Condition-Based Prognostics 2028.4.2.1 Subjective Probability 2028.4.2.2 Subjective Probabilities in Condition-Based Prognostics 2038.4.3 Why is RUL Prediction Uncertain? 2038.5 Uncertainty Quantification and Propagation for RUL Prediction 2038.5.1 Computational Framework for Uncertainty Quantification 2048.5.1.1 Present State Estimation 2048.5.1.2 Future State Prediction 2058.5.1.3 RUL Computation 2058.5.2 RUL Prediction: An Uncertainty Propagation Problem 2068.5.3 Uncertainty PropagationMethods 2068.5.3.1 Sampling-Based Methods 2078.5.3.2 AnalyticalMethods 2098.5.3.3 Hybrid Methods 2098.5.3.4 Summary of Methods 2098.6 Uncertainty Management 2108.7 Case Study: Uncertainty Quantification in the Power System of an Unmanned Aerial Vehicle 2118.7.1 Description of the Model 2118.7.2 Sources of Uncertainty 2128.7.3 Results: Constant Amplitude Loading Conditions 2138.7.4 Results: Variable Amplitude Loading Conditions 2148.7.5 Discussion 2148.8 Existing Challenges 2158.8.1 Timely Predictions 2158.8.2 Uncertainty Characterization 2168.8.3 Uncertainty Propagation 2168.8.4 Capturing Distribution Properties 2168.8.5 Accuracy 2168.8.6 Uncertainty Bounds 2168.8.7 Deterministic Calculations 2168.9 Summary 217References 2179 PHM Cost and Return on Investment 221Peter Sandborn, ChrisWilkinson, Kiri Lee Sharon, Taoufik Jazouli, and Roozbeh Bakhshi9.1 Return on Investment 2219.1.1 PHM ROI Analyses 2229.1.2 Financial Costs 2249.2 PHM Cost-Modeling Terminology and Definitions 2259.3 PHM Implementation Costs 2269.3.1 Nonrecurring Costs 2269.3.2 Recurring Costs 2279.3.3 Infrastructure Costs 2289.3.4 Nonmonetary Considerations and Maintenance Culture 2289.4 Cost Avoidance 2299.4.1 Maintenance Planning Cost Avoidance 2319.4.2 Discrete-Event Simulation Maintenance PlanningModel 2329.4.3 Fixed-Schedule Maintenance Interval 2339.4.4 Data-Driven (Precursor to Failure Monitoring) Methods 2339.4.5 Model-Based (LRU-Independent)Methods 2349.4.6 Discrete-Event Simulation Implementation Details 2369.4.7 Operational Profile 2379.5 Example PHM Cost Analysis 2389.5.1 Single-Socket Model Results 2399.5.2 Multiple-Socket Model Results 2419.6 Example Business Case Construction: Analysis for ROI 2469.7 Summary 255References 25510 Valuation and Optimization of PHM-Enabled Maintenance Decisions 261Xin Lei, Amir Reza Kashani-Pour, Peter Sandborn, and Taoufik Jazouli10.1 Valuation and Optimization of PHM-Enabled Maintenance Decisions for an Individual System 26210.1.1 A PHM-Enabled Predictive Maintenance OptimizationModel for an Individual System 26310.1.2 Case Study: Optimization of PHM-Enabled Maintenance Decisions for an Individual System (Wind Turbine) 26510.2 Availability 26810.2.1 The Business of Availability: Outcome-Based Contracts 26910.2.2 Incorporating Contract Terms into Maintenance Decisions 27010.2.3 Case Study: Optimization of PHM-Enabled Maintenance Decisions for Systems (Wind Farm) 27010.3 Future Directions 27210.3.1 Design for Availability 27210.3.2 Prognostics-BasedWarranties 27510.3.3 Contract Engineering 276References 27711 Health and Remaining Useful Life Estimation of Electronic Circuits 279Arvind Sai Sarathi Vasan and Michael G. Pecht11.1 Introduction 27911.2 RelatedWork 28111.2.1 Component-Centric Approach 28111.2.2 Circuit-Centric Approach 28211.3 Electronic Circuit Health Estimation Through Kernel Learning 28511.3.1 Kernel-Based Learning 28511.3.2 Health Estimation Method 28611.3.2.1 Likelihood-Based Function for Model Selection 28811.3.2.2 Optimization Approach for Model Selection 28911.3.3 Implementation Results 29211.3.3.1 Bandpass Filter Circuit 29311.3.3.2 DC–DC Buck Converter System 30011.4 RUL Prediction Using Model-Based Filtering 30611.4.1 Prognostics Problem Formulation 30611.4.2 Circuit DegradationModeling 30711.4.3 Model-Based Prognostic Methodology 31011.4.4 Implementation Results 31311.4.4.1 Low-Pass Filter Circuit 31311.4.4.2 Voltage Feedback Circuit 31511.4.4.3 Source of RUL Prediction Error 32011.4.4.4 Effect of First-Principles-Based Modeling 32011.5 Summary 322References 32412 PHM-Based Qualification of Electronics 329Preeti S. Chauhan12.1 Why is Product Qualification Important? 32912.2 Considerations for Product Qualification 33112.3 Review of Current Qualification Methodologies 33412.3.1 Standards-Based Qualification 33412.3.2 Knowledge-Based or PoF-Based Qualification 33712.3.3 Prognostics and Health Management-Based Qualification 34012.3.3.1 Data-Driven Techniques 34012.3.3.2 Fusion Prognostics 34312.4 Summary 345References 34613 PHM of Li-ion Batteries 349Saurabh Saxena, Yinjiao Xing, andMichael G. Pecht13.1 Introduction 34913.2 State of Charge Estimation 35113.2.1 SOC Estimation Case Study I 35213.2.1.1 NN Model 35313.2.1.2 Training and Testing Data 35413.2.1.3 Determination of the NN Structure 35513.2.1.4 Training and Testing Results 35613.2.1.5 Application of Unscented Kalman Filter 35713.2.2 SOC Estimation Case Study II 35713.2.2.1 OCV–SOC-T Test 35813.2.2.2 Battery Modeling and Parameter Identification 35913.2.2.3 OCV–SOC-T Table for Model Improvement 36013.2.2.4 Validation of the Proposed Model 36213.2.2.5 Algorithm Implementation for Online Estimation 36213.3 State of Health Estimation and Prognostics 36513.3.1 Case Study for Li-ion Battery Prognostics 36613.3.1.1 Capacity DegradationModel 36613.3.1.2 Uncertainties in Battery Prognostics 36813.3.1.3 Model Updating via Bayesian Monte Carlo 36813.3.1.4 SOH Prognostics and RUL Estimation 36913.3.1.5 Prognostic Results 37113.4 Summary 371References 37214 PHM of Light-Emitting Diodes 377Moon-Hwan Chang, Jiajie Fan, Cheng Qian, and Bo Sun14.1 Introduction 37714.2 Review of PHM Methodologies for LEDs 37814.2.1 Overview of Available Prognostic Methods 37814.2.2 Data-DrivenMethods 37914.2.2.1 Statistical Regression 37914.2.2.2 Static Bayesian Network 38114.2.2.3 Kalman Filtering 38214.2.2.4 Particle Filtering 38314.2.2.5 Artificial Neural Network 38414.2.3 Physics-Based Methods 38514.2.4 LED System-Level Prognostics 38714.3 Simulation-Based Modeling and Failure Analysis for LEDs 38814.3.1 LED Chip-LevelModeling and Failure Analysis 38914.3.1.1 Electro-optical Simulation of LED Chip 38914.3.1.2 LED Chip-Level Failure Analysis 39314.3.2 LED Package-Level Modeling and Failure Analysis 39514.3.2.1 Thermal and Optical Simulation for Phosphor-Converted White LED Package 39514.3.2.2 LED Package-Level Failure Analysis 39714.3.3 LED System-LevelModeling and Failure Analysis 39914.4 Return-on-Investment Analysis of Applying Health Monitoring to LED Lighting Systems 40114.4.1 ROI Methodology 40314.4.2 ROI Analysis of Applying System Health Monitoring to LED Lighting Systems 40614.4.2.1 Failure Rates and Distributions for ROI Simulation 40714.4.2.2 Determination of Prognostics Distance 41014.4.2.3 IPHM, CPHM, and Cu Evaluation 41214.4.2.4 ROI Evaluation 41714.5 Summary 419References 42015 PHM in Healthcare 431Mary Capelli-Schellpfeffer,Myeongsu Kang, andMichael G. Pecht15.1 Healthcare in the United States 43115.2 Considerations in Healthcare 43215.2.1 Clinical Consideration in ImplantableMedical Devices 43215.2.2 Considerations in Care Bots 43315.3 Benefits of PHM 43815.3.1 Safety Increase 43915.3.2 Operational Reliability Improvement 44015.3.3 Mission Availability Increase 44015.3.4 System’s Service Life Extension 44115.3.5 Maintenance Effectiveness Increase 44115.4 PHM of ImplantableMedical Devices 44215.5 PHM of Care Bots 44415.6 Canary-Based Prognostics of Healthcare Devices 44515.7 Summary 447References 44716 PHM of Subsea Cables 451David Flynn, Christopher Bailey, Pushpa Rajaguru,Wenshuo Tang, and Chunyan Yin16.1 Subsea Cable Market 45116.2 Subsea Cables 45216.3 Cable Failures 45416.3.1 Internal Failures 45516.3.2 Early-Stage Failures 45516.3.3 External Failures 45516.3.4 Environmental Conditions 45516.3.5 Third-Party Damage 45616.4 State-of-the-Art Monitoring 45716.5 Qualifying and Maintaining Subsea Cables 45816.5.1 Qualifying Subsea Cables 45816.5.2 Mechanical Tests 45816.5.3 Maintaining Subsea Cables 45916.6 Data-Gathering Techniques 46016.7 Measuring theWear Behavior of Cable Materials 46116.8 Predicting Cable Movement 46316.8.1 Sliding Distance Derivation 46316.8.2 Scouring Depth Calculations 46516.9 Predicting Cable Degradation 46616.9.1 Volume Loss due to Abrasion 46616.9.2 Volume Loss due to Corrosion 46616.10 Predicting Remaining Useful Life 46816.11 Case Study 47116.12 Future Challenges 47116.12.1 Data-Driven Approach for Random Failures 47116.12.2 Model-Driven Approach for Environmental Failures 47316.12.2.1 Fusion-Based PHM 47316.12.2.2 Sensing Techniques 47416.13 Summary 474References 47517 Connected Vehicle Diagnostics and Prognostics 479Yilu Zhang and Xinyu Du17.1 Introduction 47917.2 Design of an Automatic Field Data Analyzer 48117.2.1 Data Collection Subsystem 48217.2.2 Information Abstraction Subsystem 48217.2.3 Root Cause Analysis Subsystem 48217.2.3.1 Feature-Ranking Module 48217.2.3.2 Relevant Feature Set Selection 48417.2.3.3 Results Interpretation 48617.3 Case Study: CVDP for Vehicle Batteries 48617.3.1 Brief Background of Vehicle Batteries 48617.3.2 Applying AFDA for Vehicle Batteries 48817.3.3 Experimental Results 489Contents xvii17.3.3.1 Information Abstraction 49017.3.3.2 Feature Ranking 49017.3.3.3 Interpretation of Results 49517.4 Summary 498References 49918 The Role of PHM at Commercial Airlines 503RhondaWalthall and Ravi Rajamani18.1 Evolution of Aviation Maintenance 50318.2 Stakeholder Expectations for PHM 50618.2.1 Passenger Expectations 50618.2.2 Airline/Operator/Owner Expectations 50718.2.3 Airframe Manufacturer Expectations 50918.2.4 Engine Manufacturer Expectations 51018.2.5 System and Component Supplier Expectations 51118.2.6 MRO Organization Expectations 51218.3 PHM Implementation 51318.3.1 SATAA 51318.4 PHM Applications 51718.4.1 Engine Health Management (EHM) 51718.4.1.1 History of EHM 51818.4.1.2 EHM Infrastructure 51918.4.1.3 Technologies Associated with EHM 52018.4.1.4 The Future 52318.4.2 Auxiliary Power Unit (APU) Health Management 52418.4.3 Environmental Control System (ECS) and Air Distribution Health Monitoring 52518.4.4 Landing System Health Monitoring 52618.4.5 Liquid Cooling System Health Monitoring 52618.4.6 Nitrogen Generation System (NGS) Health Monitoring 52718.4.7 Fuel Consumption Monitoring 52718.4.8 Flight Control Actuation Health Monitoring 52818.4.9 Electric Power System Health Monitoring 52918.4.10 Structural Health Monitoring (SHM) 52918.4.11 Battery Health Management 53118.5 Summary 532References 53319 PHM Software for Electronics 535Noel Jordan Jameson,Myeongsu Kang, and Jing Tian19.1 PHM Software: CALCE Simulation Assisted Reliability Assessment 53519.2 PHM Software: Data-Driven 54019.2.1 Data Flow 54119.2.2 Master Options 54219.2.3 Data Pre-processing 54319.2.4 Feature Discovery 54519.2.5 Anomaly Detection 54619.2.6 Diagnostics/Classification 54819.2.7 Prognostics/Modeling 55219.2.8 Challenges in Data-Driven PHM Software Development 55419.3 Summary 55720 eMaintenance 559Ramin Karim, Phillip Tretten, and Uday Kumar20.1 From Reactive to Proactive Maintenance 55920.2 The Onset of eMaintenance 56020.3 MaintenanceManagement System 56120.3.1 Life-cycle Management 56220.3.2 eMaintenance Architecture 56420.4 Sensor Systems 56420.4.1 Sensor Technology for PHM 56520.5 Data Analysis 56520.6 Predictive Maintenance 56620.7 Maintenance Analytics 56720.7.1 Maintenance Descriptive Analytics 56820.7.2 Maintenance Analytics and eMaintenance 56820.7.3 Maintenance Analytics and Big Data 56820.8 Knowledge Discovery 57020.9 Integrated Knowledge Discovery 57120.10 User Interface for Decision Support 57220.11 Applications of eMaintenance 57220.11.1 eMaintenance in Railways 57220.11.1.1 Railway Cloud: Swedish Railway Data 57320.11.1.2 Railway Cloud: Service Architecture 57320.11.1.3 Railway Cloud: Usage Scenario 57420.11.2 eMaintenance in Manufacturing 57420.11.3 MEMS Sensors for Bearing Vibration Measurement 57620.11.4 Wireless Sensors for Temperature Measurement 57620.11.5 Monitoring Systems 57620.11.6 eMaintenance Cloud and Servers 57820.11.7 Dashboard Managers 58020.11.8 Alarm Servers 58020.11.9 Cloud Services 58120.11.10 Graphic User Interfaces 58320.12 Internet Technology and Optimizing Technology 585References 58621 Predictive Maintenance in the IoT Era 589Rashmi B. Shetty21.1 Background 58921.1.1 Challenges of a Maintenance Program 59021.1.2 Evolution of Maintenance Paradigms 59021.1.3 Preventive Versus Predictive Maintenance 59221.1.4 P–F Curve 59221.1.5 Bathtub Curve 59421.2 Benefits of a Predictive Maintenance Program 59521.3 Prognostic Model Selection for Predictive Maintenance 59621.4 Internet ofThings 59821.4.1 Industrial IoT 59821.5 Predictive Maintenance Based on IoT 59921.6 Predictive Maintenance Usage Cases 60021.7 Machine Learning Techniques for Data-Driven Predictive Maintenance 60021.7.1 Supervised Learning 60221.7.2 Unsupervised Learning 60221.7.3 Anomaly Detection 60221.7.4 Multi-class and Binary Classification Models 60321.7.5 Regression Models 60421.7.6 Survival Models 60421.8 Best Practices 60421.8.1 Define Business Problem and QuantitativeMetrics 60521.8.2 Identify Assets and Data Sources 60521.8.3 Data Acquisition and Transformation 60621.8.4 Build Models 60721.8.5 Model Selection 60721.8.6 Predict Outcomes and Transform into Process Insights 60821.8.7 Operationalize and Deploy 60921.8.8 Continuous Monitoring 60921.9 Challenges in a Successful Predictive Maintenance Program 61021.9.1 Predictive Maintenance Management Success Key Performance Indicators (KPIs) 61021.10 Summary 611References 61122 Analysis of PHM Patents for Electronics 613Zhenbao Liu, Zhen Jia, Chi-Man Vong, Shuhui Bu, andMichael G. Pecht22.1 Introduction 61322.2 Analysis of PHM Patents for Electronics 61622.2.1 Sources of PHM Patents 61622.2.2 Analysis of PHM Patents 61722.3 Trend of Electronics PHM 61922.3.1 Semiconductor Products and Computers 61922.3.2 Batteries 62222.3.3 Electric Motors 62622.3.4 Circuits and Systems 62922.3.5 Electrical Devices in Automobiles and Airplanes 63122.3.6 Networks and Communication Facilities 63422.3.7 Others 63622.4 Summary 638References 63923 A PHM Roadmap for Electronics-Rich Systems 64Michael G. Pecht23.1 Introduction 64923.2 Roadmap Classifications 65023.2.1 PHM at the Component Level 65123.2.1.1 PHM for Integrated Circuits 65223.2.1.2 High-Power Switching Electronics 65223.2.1.3 Built-In Prognostics for Components and Circuit Boards 65323.2.1.4 Photo-Electronics Prognostics 65423.2.1.5 Interconnect andWiring Prognostics 65623.2.2 PHM at the System Level 65723.2.2.1 Legacy Systems 65723.2.2.2 Environmental and OperationalMonitoring 65923.2.2.3 LRU to Device Level 65923.2.2.4 Dynamic Reconfiguration 65923.2.2.5 System Power Management and PHM 66023.2.2.6 PHM as Knowledge Infrastructure for System Development 66023.2.2.7 Prognostics for Software 66023.2.2.8 PHM for Mitigation of Reliability and Safety Risks 66123.2.2.9 PHM in Supply Chain Management and Product Maintenance 66223.3 Methodology Development 66323.3.1 Best Algorithms 66423.3.1.1 Approaches to Training 66723.3.1.2 Active Learning for Unlabeled Data 66723.3.1.3 Sampling Techniques and Cost-Sensitive Learning for Imbalanced Data 66823.3.1.4 Transfer Learning for Knowledge Transfer 66823.3.1.5 Internet ofThings and Big Data Analytics 66923.3.2 Verification and Validation 67023.3.3 Long-Term PHM Studies 67123.3.4 PHM for Storage 67123.3.5 PHM for No-Fault-Found/Intermittent Failures 67223.3.6 PHM for Products Subjected to Indeterminate Operating Conditions 67323.4 Nontechnical Barriers 67423.4.1 Cost, Return on Investment, and Business Case Development 67423.4.2 Liability and Litigation 67623.4.2.1 Code Architecture: Proprietary or Open? 67623.4.2.2 Long-Term Code Maintenance and Upgrades 67623.4.2.3 False Alarms, Missed Alarms, and Life-Safety Implications 67723.4.2.4 Warranty Restructuring 67723.4.3 Maintenance Culture 67723.4.4 Contract Structure 67723.4.5 Role of Standards Organizations 67823.4.5.1 IEEE Reliability Society and PHM Efforts 67823.4.5.2 SAE PHM Standards 67823.4.5.3 PHM Society 67923.4.6 Licensing and Entitlement Management 680References 680Appendix A Commercially Available Sensor Systems for PHM 691A.1 SmartButton – ACR Systems 691A.2 OWL 400 – ACR Systems 693A.3 SAVERTM 3X90 – Lansmont Instruments 695A.4 G-Link®-LXRS®– LORD MicroStrain®Sensing Systems 697A.5 V-Link®-LXRS®– LORD MicroStrain Sensing Systems 699A.6 3DM-GX4–25TM – LORD MicroStrain Sensing Systems 702A.7 IEPE-LinkTM-LXRS®– LORD MicroStrain Sensing Systems 704A.8 ICHM®20/20 – Oceana Sensor 706A.9 EnvironmentalMonitoring System 200TM – Upsite Technologies 708A.10 S2NAP®– RLWInc. 710A.11 SR1 Strain Gage Indicator – Advance Instrument Inc. 712A.12 P3 Strain Indicator and Recorder – Micro-Measurements 714A.13 Airscale Suspension-BasedWeighing System – VPG Inc. 716A.14 Radio Microlog – Transmission Dynamics 718Appendix B Journals and Conference Proceedings Related to PHM 721B.1 Journals 721B.2 Conference Proceedings 722Appendix C Glossary of Terms and Definitions 725Index 731