E-Book, Englisch, 549 Seiten
Galar Pascual Artificial Intelligence Tools
Erscheinungsjahr 2015
ISBN: 978-1-4665-8406-8
Verlag: Taylor & Francis
Format: PDF
Kopierschutz: Adobe DRM (»Systemvoraussetzungen)
Decision Support Systems in Condition Monitoring and DIagnosis
E-Book, Englisch, 549 Seiten
ISBN: 978-1-4665-8406-8
Verlag: Taylor & Francis
Format: PDF
Kopierschutz: Adobe DRM (»Systemvoraussetzungen)
Artificial intelligence (AI) represents a programming approach and methodology that has evolved and is still evolving for effective maintenance management through its use in condition monitoring (CM). The most prevalent application of AI has been fault diagnosis and trouble shooting in maintenance of industrial machinery. AI tools are used in maintenance of all types of industrial machinery. Due to the success of AI in CM, this book compiles all AI tools used in CMg to structure them in a reference handbook.
Autoren/Hrsg.
Weitere Infos & Material
Massive field data collection: Issues and challenges
Maintenance data: different sources and disparate nature
Required data for diagnosis and prognosis
Definition of representative normal regions
The boundary between normal and outlying behaviour
Notion of an outlier for different application domains
Availability of labelled data for training/validation
Malicious adversaries
Data and noise
Normal behaviour and its evolution
Condition monitoring: Available techniques
The role of CM in CBM and PdM
Difference between CM and NDT
Oil Analysis
Vibration Analysis
Motor Circuit Analysis
Thermography
Ultrasonic Monitoring
Performance Monitoring through automation data, process data and other information
sources
Challenges of Condition Monitoring using AI techniques
Anomaly detection
Rare Class Mining
Chance discovery
Novelty Detection
Exception Mining
Noise Removal
The Swan Song
Input and output data
Supervised Failure Detection
Semi-supervised Failure Detection
Unsupervised Failure Detection
Individual Failures
Contextual Failures
Collective Failures
Classification Based Techniques
Neural network based approaches
Support Vector machines (SVM) based approaches
Bayesian networks based approaches
Liquid States Machines (LSM) and other Reservoir Computing methods
Nearest Neighbor Based Techniques
The concept of neighborhood
Distance based methods
Density based methods
Local Outlier Factor (LOF)
Connectivity Outlier Factor (COF)
Multi-Granularity Deviation Factor (MDEF)
The use of NNB in semi supervised and unsupervised learning
Clustering Based Techniques
Categorization: Semi-Supervised and unsupervised
Data records with problems to fit into any cluster (residuals from clustering)
The problem of small clusters
Low density clusters or local anomalies
Computational cost and problems of outliers in clusters
Statistical techniques
The use of stochastic distributions to detect outliers
Issues related to data set size
Parametric Techniques
Non-parametric Techniques
Information Theory Based Techniques
Information contained in the maintenance data
Entropy and relative entropy estimation
The detection of alterations in information content
Estimation of optimum size of data set
The advantages of information theory as unsupervised system
The uncertainty management
Classical logic and fuzzy logic
Using fuzzy logic to solve diagnosis problems
Defuzzification
The need for complex relations in contextual decision making
Bayesian analysis versus classical statistical analysis
