
Industrial Fault Diagnosis and Remaining Useful Life Prediction
Cross-Domain, Zero-Sample, and Degradation Modeling Methods
- 1st Edition - February 1, 2026
- Imprint: Elsevier
- Authors: Hongpeng Yin, Li Cai, Peng Zhang
- Language: English
- Paperback ISBN:9 7 8 - 0 - 4 4 3 - 4 4 2 9 1 - 9
- eBook ISBN:9 7 8 - 0 - 4 4 3 - 4 4 2 9 2 - 6
Industrial Fault Diagnosis and Remaining Useful Life Prediction introduces zero-sample learning methods that enable fault diagnosis and Predict Remaining Useful Life (RUL) withou… Read more
Purchase options

Industrial Fault Diagnosis and Remaining Useful Life Prediction introduces zero-sample learning methods that enable fault diagnosis and Predict Remaining Useful Life (RUL) without the need for labelled fault data. This is particularly valuable in industrial settings where labelled data is scarce or unavailable. Offers step-by-step guidance on implementing zero-shot learning models using real industrial data, reducing the learning curve for practitioners; includes real-world industrial case studies to demonstrate the application of zero-sample learning techniques in various industries, such as manufacturing, energy, and transportation. Such case studies provide readers with actionable insights and practical solutions. The book covers advanced methodologies for predicting the remaining useful life of industrial equipment, supporting readers in optimizing maintenance schedules, reducing downtime and extending the lifespan of critical assets. Covers state-of-the-art algorithms, including deep learning, transfer learning and domain adaptation, tailored for zero-sample scenarios. These tools empower readers to develop robust fault diagnosis and RUL prediction systems, enhancing predictive maintenance capabilities and ensuring the reliability of industrial systems
- Introduces zero-shot learning techniques that enable fault diagnosis and Remaining Useful Life or RUL prediction even with limited or no labelled data for specific faults
- Provides methodologies for models to generalize for unseen faults, ensuring robust performance in real-world scenarios
- Offers step-by-step guidance on implementing zero-shot learning models using real industrial data, reducing the learning curve for practitioners and the ability to implement advanced techniques: thereby enhancing predictive maintenance capabilities and ensuring the reliability of industrial systems
- Includes real-world case studies and examples to demonstrate the application of zero-shot learning in industrial settings, bridging the gap between theory and practice
Industrial engineers and researchers engaged in equipment fault diagnosis and lifespan prediction research
1: Introduction
1.1 Intelligent fault diagnosis
1.1.1 Cross-domain fault diagnosis
1.1.2 Zero-sample fault diagnosis
1.1.3 Federated fault diagnosis
1.2 Degradation modeling and intelligent prognosis
1.2.1 Statistics-based prognosis
1.2.2 Time series analysis-based prognosis
1.2.3 Artificial intelligence-based prognosis References
2: Basic theories and methods of intelligent fault diagnosis and health prediction
2.1 Advanced fault diagnosis methods
2.1.1 Embedded industrial zero-shot learning model
2.1.1.1 Mapping from data space to semantic space
2.1.1.2 Mapping from semantic space to data space
2.1.2 Generative industrial zero-shot learning models
2.1.2.1 Generation based on variational autoencoder
2.1.2.2 Generation based on generative adversarial network
2.2 Advanced health prognosis methods
2.2.1 Remaining useful life (RUL) prediction method based on PSO
2.2.2 RUL prediction method based on Wiener process
2.2.3 RUL prediction method based on GPR
2.2.4 RUL prediction method based on HMM
2.2.4.1 Degradation factor design
2.2.4.2 Degradation transition probability calculation References
3: Multi-attribute learning framework for zero-sample fault detection in machinery
3.1 Motivation
3.2 Related works
3.3 Methodology
3.3.1 Problem Statement
3.3.2 The framework of zero-sample fault diagnosis
3.3.3 Data collecting and variables selecting
3.3.4 Data-to-image transformation process
3.3.5 CNN-based feature extraction and classifier design
3.3.6 Zero-sample fault learning and diagnosis
3.4 Case study
3.4.1 CWRU rolling bearing fault case
3.4.2 KAT bearing fault case
3.5 Conclusions References
4: Generalized zero-sample industrial fault diagnosis with domain bias
4.1 Motivation
4.2 Related works
4.2.1 Generalized zero-shot learning
4.2.2 Domain bias problem in GZSL
4.3 Methodology
4.3.1 Problem statement
4.3.2 Generalized zero-sample fault diagnosis
4.3.3 ResNet-based 1D CNN for feature extraction
4.3.4 Unseen fault detection for GZSFD
4.3.5 Clustering for the detected unseen class
4.3.6 Semantic embedding strategies for the detected seen class
4.4 Case study
4.4.1 CWRU rolling bearing fault case
4.4.2 Industrial three-phase flow process case
4.5 Conclusions References
5: Generalized zero-sample industrial fault diagnosis under cross-domain scenarios
5.1 Motivation
5.2 Problem definition and notations
5.3 Methodology
5.3.1 Generalized zero-sample fault diagnosis
5.3.2 Feature extraction and feature alignment
5.3.3 Unseen fault detection for GZSFD
5.3.4 Fault diagnosis procedure
5.4 Case study
5.4.1 CWRU rolling bearing fault case
5.4.2 Industrial three-phase flow process case
5.5 Conclusions References
6: Learning across multisource domains for generalized zero-sample industrial fault diagnosis
6.1 Motivation
6.2 Notations and problem formulation
6.3 Methodology
6.3.1 Proposed framework
6.3.2 Learning across multisource domains
6.3.3 Mappings learning
6.3.4 Procedure for generalized zero-sample fault diagnosis
6.4 Case study
6.4.1 CWRU rolling bearing fault case
6.4.2 Industrial three-phase flow process case
6.5 Conclusions References
7: Federated generalized zero-sample industrial fault diagnosis across multisource domains
7.1 Motivation
7.2 Preliminaries
7.2.1 Stacked autoencoder
7.2.2 Maximum mean discrepancy
7.3 Methodology
7.3.1 Notations and problem formulation
7.3.2 Proposed framework
7.3.3 Feature extraction and feature alignment
7.3.4 Bidirectional autoencoder
7.3.5 Gating mechanism based on BAE
7.3.6 Procedure for federated GZSFD
7.4 Case study
7.4.1 CWRU rolling bearing fault case
7.4.2 Industrial three-phase flow process case
7.5 Conclusions References
8: A multi-phase Wiener process-based degradation model with imperfect maintenance activities
8.1 Motivation
8.2 Preliminaries
8.3 Methodology
8.3.1 Prior specification
8.3.2 Residual damage parameters estimation
8.3.3 Degradation parameters estimation
8.4 Model updating and RUL distribution
8.4.1 Model updating
8.4.2 RUL distribution
8.5 Case study
8.5.1 Numerical study
8.5.2 Practical study
8.6 Conclusions References
1.1 Intelligent fault diagnosis
1.1.1 Cross-domain fault diagnosis
1.1.2 Zero-sample fault diagnosis
1.1.3 Federated fault diagnosis
1.2 Degradation modeling and intelligent prognosis
1.2.1 Statistics-based prognosis
1.2.2 Time series analysis-based prognosis
1.2.3 Artificial intelligence-based prognosis References
2: Basic theories and methods of intelligent fault diagnosis and health prediction
2.1 Advanced fault diagnosis methods
2.1.1 Embedded industrial zero-shot learning model
2.1.1.1 Mapping from data space to semantic space
2.1.1.2 Mapping from semantic space to data space
2.1.2 Generative industrial zero-shot learning models
2.1.2.1 Generation based on variational autoencoder
2.1.2.2 Generation based on generative adversarial network
2.2 Advanced health prognosis methods
2.2.1 Remaining useful life (RUL) prediction method based on PSO
2.2.2 RUL prediction method based on Wiener process
2.2.3 RUL prediction method based on GPR
2.2.4 RUL prediction method based on HMM
2.2.4.1 Degradation factor design
2.2.4.2 Degradation transition probability calculation References
3: Multi-attribute learning framework for zero-sample fault detection in machinery
3.1 Motivation
3.2 Related works
3.3 Methodology
3.3.1 Problem Statement
3.3.2 The framework of zero-sample fault diagnosis
3.3.3 Data collecting and variables selecting
3.3.4 Data-to-image transformation process
3.3.5 CNN-based feature extraction and classifier design
3.3.6 Zero-sample fault learning and diagnosis
3.4 Case study
3.4.1 CWRU rolling bearing fault case
3.4.2 KAT bearing fault case
3.5 Conclusions References
4: Generalized zero-sample industrial fault diagnosis with domain bias
4.1 Motivation
4.2 Related works
4.2.1 Generalized zero-shot learning
4.2.2 Domain bias problem in GZSL
4.3 Methodology
4.3.1 Problem statement
4.3.2 Generalized zero-sample fault diagnosis
4.3.3 ResNet-based 1D CNN for feature extraction
4.3.4 Unseen fault detection for GZSFD
4.3.5 Clustering for the detected unseen class
4.3.6 Semantic embedding strategies for the detected seen class
4.4 Case study
4.4.1 CWRU rolling bearing fault case
4.4.2 Industrial three-phase flow process case
4.5 Conclusions References
5: Generalized zero-sample industrial fault diagnosis under cross-domain scenarios
5.1 Motivation
5.2 Problem definition and notations
5.3 Methodology
5.3.1 Generalized zero-sample fault diagnosis
5.3.2 Feature extraction and feature alignment
5.3.3 Unseen fault detection for GZSFD
5.3.4 Fault diagnosis procedure
5.4 Case study
5.4.1 CWRU rolling bearing fault case
5.4.2 Industrial three-phase flow process case
5.5 Conclusions References
6: Learning across multisource domains for generalized zero-sample industrial fault diagnosis
6.1 Motivation
6.2 Notations and problem formulation
6.3 Methodology
6.3.1 Proposed framework
6.3.2 Learning across multisource domains
6.3.3 Mappings learning
6.3.4 Procedure for generalized zero-sample fault diagnosis
6.4 Case study
6.4.1 CWRU rolling bearing fault case
6.4.2 Industrial three-phase flow process case
6.5 Conclusions References
7: Federated generalized zero-sample industrial fault diagnosis across multisource domains
7.1 Motivation
7.2 Preliminaries
7.2.1 Stacked autoencoder
7.2.2 Maximum mean discrepancy
7.3 Methodology
7.3.1 Notations and problem formulation
7.3.2 Proposed framework
7.3.3 Feature extraction and feature alignment
7.3.4 Bidirectional autoencoder
7.3.5 Gating mechanism based on BAE
7.3.6 Procedure for federated GZSFD
7.4 Case study
7.4.1 CWRU rolling bearing fault case
7.4.2 Industrial three-phase flow process case
7.5 Conclusions References
8: A multi-phase Wiener process-based degradation model with imperfect maintenance activities
8.1 Motivation
8.2 Preliminaries
8.3 Methodology
8.3.1 Prior specification
8.3.2 Residual damage parameters estimation
8.3.3 Degradation parameters estimation
8.4 Model updating and RUL distribution
8.4.1 Model updating
8.4.2 RUL distribution
8.5 Case study
8.5.1 Numerical study
8.5.2 Practical study
8.6 Conclusions References
- Edition: 1
- Published: February 1, 2026
- Imprint: Elsevier
- Language: English
HY
Hongpeng Yin
Professor Hongpeng Yin is based at the School of Automation, Chongqing University in China. His current research interests mainly include data-driven process monitoring and fault diagnosis, pattern recognition, and data mining
Affiliations and expertise
Chongqing University, ChinaLC
Li Cai
Li Cai received the B.E. degree from the School of Physics and Electronic Engineering from Hainan Normal University in 2019. He is currently undertaking a Ph.D. degree at the School of Automation, Chongqing University, China. His major research interests include data-driven fault detection and diagnosis, fault prediction, remaining useful life prediction, and (generalized) zero-shot learning
Affiliations and expertise
Chongqing University, ChinaPZ
Peng Zhang
Peng Zhang received the B.E. degree from College of Automation, Hangzhou Dianzi University, China in 2021. He is currently working towards a Ph.D. degree in the College of Automation, Chongqing University, China. His research interests include data mining, fault diagnosis and machine learning
Affiliations and expertise
Chongqing University, China