Machine Learning and Knowledge Discovery for Engineering Systems Health Management

CHAPTER 3

Anomaly Detection in a Fleet of Systems

Nikunj OzaSantanu Das

CONTENTS

3.1    What Is a Fleet?

3.2    Background

3.2.1    Past Work in Fleet Health Monitoring

3.2.1.1    Numeric Data

3.2.1.2    Text

3.2.2    Role of Data Mining in Fleet Health Monitoring

3.2.2.1    Anomaly Detection Problem

3.2.2.2    Some Preliminaries

3.3    Key Issues in Fleet Health Monitoring

3.3.1    Heterogeneity

3.3.2    Efficiency: Large Data Volume, Distributed Data

3.4    Dealing with Heterogeneity

3.4.1    Orca

3.4.2    GritBot

3.4.3    Kernel-Based Anomaly Detection

3.4.3.1    Kernel Theory and Operations

3.4.3.2    Is It a Mercer Kernel?

3.4.3.3    Problem-Specific Kernel Functions

3.4.3.4    Information Fusion

3.4.3.5    One-Class SVMs: An Overview ...