Artificial Intelligence Tools: Decision Support Systems in Condition Monitoring and DIagnosis, 1st Edition (Hardback) book cover

Artificial Intelligence Tools

Decision Support Systems in Condition Monitoring and DIagnosis, 1st Edition

By Diego Galar Pascual

CRC Press

549 pages | 188 B/W Illus.

Purchasing Options:$ = USD
Hardback: 9781466584051
pub: 2015-04-22
$195.00
x
eBook (VitalSource) : 9780429102325
pub: 2015-04-22
from $28.98


FREE Standard Shipping!

Description

Artificial Intelligence Tools: Decision Support Systems in Condition Monitoring and Diagnosis discusses various white- and black-box approaches to fault diagnosis in condition monitoring (CM). This indispensable resource:

  • Addresses nearest-neighbor-based, clustering-based, statistical, and information theory-based techniques
  • Considers the merits of each technique as well as the issues associated with real-life application
  • Covers classification methods, from neural networks to Bayesian and support vector machines
  • Proposes fuzzy logic to explain the uncertainties associated with diagnostic processes
  • Provides data sets, sample signals, and MATLAB® code for algorithm testing

Artificial Intelligence Tools: Decision Support Systems in Condition Monitoring and Diagnosis delivers a thorough evaluation of the latest AI tools for CM, describing the most common fault diagnosis techniques used and the data acquired when these techniques are applied.

Reviews

"… a long overdue publication; the condition monitoring community, from newcomers to experts, will find themselves constantly referring to this book, especially to find definitive answers to often debated issues."

—Chris Pomfret, Society for Machinery Failure Prevention Technology, Dayton, Ohio, USA

"… a good reference book for students, educators, and maintenance engineers who would like to use artificial intelligence (AI) techniques for data fusion and decision making in condition monitoring and diagnosis."

—Zhongxiao Peng, University of New South Wales, Sydney, Australia

"… a detailed and descriptive analysis of the latest thinking on data collection and analyses for maintenance task development. … an important addition to the library of existing knowledge to support asset managers, academics, and engineering students who want to understand the methods and techniques to diagnose the state of an asset and develop a new approach to asset management."

—David Baglee, University of Sunderland, UK

"… very comprehensive and informative in its coverage of condition monitoring and condition-based maintenance for machinery. I’m not aware of any other book on the market that has the breadth of coverage of this book. It will be an excellent resource for practitioners in the field. The book contains well-written and very understandable definitions and descriptions of the techniques used for condition monitoring for machinery, providing a useful resource for students and practicing engineers."

—Peter Sandborn, University of Maryland, College Park, USA

Table of Contents

Preface

Acknowledgments

Author

Massive Field Data Collection: Issues and Challenges

An Introduction to Systems

Evolution of Mathematical Models

Models as Approximations of Reality

Modeling Classification Based on Purpose

Model Construction

Modeling and Simulation

Modeling and Simulation Process

Simulation Model

System Identification Problem

Key Features of the Identification Problem

Identification Steps

Identifiability

Classes of Models for Identification

Introduction to the Concept of Diagnostics

Meaning and Impact of Diagnostics

Concepts, Methods, and Techniques of Diagnostics

Application of Technical Diagnostics

Management of Failure Analysis

Process of Diagnosis

The History of Diagnosis

Big Data in Maintenance

Maintenance Data: Different Sources and Disparate Nature

Required Data for Diagnosis and Prognosis

Existing Data in the Maintenance Function

In Search of a Comprehensive Data Format

Database Structure

CM Data and Automatic Asset Data Collection

References

Condition Monitoring: Available Techniques

Role of Condition Monitoring in Condition-Based Maintenance and Predictive Maintenance

Difference between CM and Nondestructive Testing

What is NDT?

Concerns about NDT

Conditions for Effective NDT

Qualification as a Main Difference

Oil Analysis

Looking Inside

Physical Tests

Metal Tests

Oil Analysis Benefits

Vibration Analysis

Machine Vibration Causes

How is Machine Vibration Described?

Vibration Sensors–Transducers

Mounting Techniques

VA Applications

Vibration Meters

Vibration Analyzer

Periodic Monitoring or Online Monitoring

Motor Circuit Analysis

MCA Application

Thermography

Contact Temperature Measurement

Noncontact Thermal Measurement

Infrared Inspection Techniques

IT Program

Applications

Image Analysis

Severity Criteria

Acoustic Technology: Sonic and Ultrasonic Monitoring

Acoustic Leak Detection

AE Crack Detection

Performance Monitoring Using Automation Data, Process Data, and Other Information Sources

Data Fusion: A Requirement in the Maintenance of Processes

XML: Protocol for Understanding Each Other

XML: Protocol to Destroy Communication Barriers

Example of Asset Data Integration Using XML

References

Challenges of Condition Monitoring Using AI Techniques

Anomaly Detection

What are Anomalies?

Challenges

Types of Anomaly

Point Anomalies

Contextual Anomalies

Collective Anomalies

Data Labels

Anomaly Detection Output

Industrial Damage Detection

Rare Class Mining

Why Rare Cases are Problematic

Methods for Rare Class Problems

Chance Discovery

Prediction Methods for Rare Events

Chance Discoveries and Data Mining

Novelty Detection

Outlier Detection

Novelty Detection Using Supervised Neural Networks

Other Models of Novelty Detection

Exception Mining

Confidence-Based Interestingness

Support-Based Interestingness

Comparison with Exception-Mining Model

Digging Out the Exceptions

Noise Removal

Distance-Based Outlier Detection Methods for Noise Removal

Density-Based Outlier Detection Method for Noise Removal

Clustering-Based Outlier Detection Methods for Noise Removal

The Black Swan

Combining Statistics and Events

The Data

Extraction

Detection of Outliers

Association Rule Mining

References

Input and Output Data

Supervised Failure Detection

Decision Trees

Bayesian Network Classifiers

Markov Models

Conditional Random Fields

Support Vector Machines

k-NN Algorithms

Semisupervised Failure Detection

Expectation-Maximization with Generative Mixture Models

Self-Training

Cotraining

Transductive SVMs

Graph-Based Methods

Unsupervised Failure Detection

Clustering

Self-Organizing Map

Adaptive Resonance Theory

Other Unsupervised Machine-Learning Algorithms

Individual Failures

Classification-Based Techniques

Nearest Neighbor-Based Techniques

Clustering-Based Techniques

Anomaly Detection Techniques Based on Statistical Approach

Other Detection Techniques

Contextual Failures

Computational Complexity

Collective Failures

Fault Detection in Mechanical Units

Structural Defect Detection

References

Two-Stage Response Surface Approaches to Modeling Drug Interaction

Classification-Based Techniques

Neural Network-Based Approaches

Supervised Networks for Classification

Unsupervised Learning

SVM-Based Approaches

Statistical-Learning Theory

SVM Representation

Kernel Trick

SVM for Classification

Strength and Weakness of SVM

Bayesian Networks-Based Approaches

Bayesian Decision Theory—Continuous Features

Minimum-Error-Rate Classification

Bayesian Classifiers

Bayesian Estimation

Liquid State Machines and Other Reservoir Computing Methods

References

Nearest-Neighbor-Based Techniques

The Concept of Neighborhood

Distance-Based Methods

Cell-Based Methods

Index-Based Methods

Reverse NN Approach

Density-Based Methods

Local Outlier Factor

Multigranularity Deviation Factor

The Use of Neural Network Based in Semisupervised and Unsupervised Learning

Semisupervised Learning with Neural Networks

Unsupervised Learning with Neural Networks

References

Clustering-Based Techniques

Categorization versus Classification

Complex Data in Maintenance: Challenges or Problems?

Introduction

What is Clustering?

The Goal of Clustering

Clustering as an Unsupervised Classification

Clustering Algorithms

Categorization: Semisupervised and Unsupervised

Unsupervised Clustering

Supervised Clustering Algorithms

Issues Using Cluster Analysis

Clustering Methods and Their Issues

Limitations of Associating Distances

Text Clustering and Categorization

Applications and Performance

Contextual Clustering

Overview of Contextual Content Analysis

Meanings in Text

Measuring Context

Issues of Validity and Reliability

References

Statistical Techniques

The Use of Stochastic Distributions to Detect Outliers

Taxonomy of Outlier Detection Methods

Univariate Statistical Methods

Multivariate Outlier Detection

Issues Related to Data Set Size

Data Size Characteristics

Small and Big Data

Big Data

Parametric Techniques

Statistical Approach

Control Chart Technique

Deviation-Based Approach

Nonparametric Techniques

Linear Regression Technique

Manhattan Distance Technique

Data-Mining Methods for Outlier Detection

References

Information Theory-Based Techniques

Introduction

Informational Universe—Pan-Informationalism

Information as a Structure: Data–Information–Knowledge–Wisdom–Weaving

Different Schools of Information

Theories of Information

Semantic Theories of Information

What is the Difference that Makes a Difference? Syntactic versus Semantic Information

No Information without Representation! Correspondence Models versus Interactive Representation

Information Contained in Maintenance Data

Feature Selection

Feature Extraction

Contextual Information

Context as Complex Information Content in Maintenance: an Example of Health Assessment for Machine Tools

Entropy and Relative Entropy Estimation

Entropy Estimation

Entropy Statistics

Applications

Detection of Alterations in Information Content

What is Information Content of an Object?

Entropy as a Measure of Information Integrity

Advantages of Information Theory as an Unsupervised System

Information and Learning Process

References

Uncertainty Management

Classical Logic and Fuzzy Logic

Classical Logic

Fuzzy Logic

The Fuzzy Set Concept

Fuzzy Networks

Using Fuzzy Logic to Solve Diagnosis Problems

Architecture for Fault Detection and Diagnosis

A Fuzzy Filter for Residual Evaluation

Identification by Fuzzy Clustering

Defuzzification

Lambda Cuts for Fuzzy Sets

Lambda Cuts for Fuzzy Relations

Defuzzification Methods

The Need for Complex Relations in Contextual Decision Making

When Do We Need Fuzzy Systems?

Applying Expert-Based Fuzzy Systems

Applying Data-Based Fuzzy Systems

Bayesian Analysis versus Classical Statistical Analysis

References

Index

About the Author

Diego Galar Pascual holds an M.Sc and Ph.D from Saragossa University, Zaragoza, Spain. He has been a professor at several universities, including Saragossa University and the European University of Madrid, Spain. At Saragossa University, he also served as director of academic innovation, director of international relations, pro-vice-chancellor, and senior researcher in the Aragon Institute of Engineering Research (i3A). In addition, he has been the technological director and CBM manager of international firms such as Volvo, Saab, Boliden, Scania, Tetrapak, Heinz, and Atlas Copco. Currently, he is the professor of condition monitoring in the Division of Operation and Maintenance of the Luleå University of Technology (LTU), Sweden, where he also is involved with the LTU-SKF University Technology Center. Widely published, Dr. Galar Pascual serves as a visiting professor at the University of Valencia (Spain), Polytechnic of Braganza (Portugal), Valley University (Mexico), Sunderland University (UK), University of Maryland (College Park, USA), and Northern Illinois University (DeKalb, USA).

Subject Categories

BISAC Subject Codes/Headings:
COM037000
COMPUTERS / Machine Theory
TEC016000
TECHNOLOGY & ENGINEERING / Industrial Design / General
TEC032000
TECHNOLOGY & ENGINEERING / Quality Control