Prognostics and Health Management

Preface

A prognostics and health management/monitoring (PHM) system can be thought of as consisting of three major systems: a sensing system consisting of a sensor and a feature vector framework, a prognosis system comprising a prediction framework and a performance‐validation framework, and a health‐management framework. Although health management is probably the most complex and most expensive, this book presents topics related to sensing systems and prognosis. An important goal of those systems is to provide accurate prognostic information regarding the prognosis of the health of the system being monitored. This book begins by presenting approaches to reliability predictions based on traditional model‐driven, data‐driven, and hybrid‐driven approaches as necessary background to understanding the rationale for the signature‐driven approaches presented later in the book. Those traditional, or handbook, methods are evaluated as inaccurate and misleading when used for prognostic estimation of a future failure.

This book then develops approaches to modeling and data handling that take into account failure modes and operational environment and conditions, and presents an approach using signatures created by extracting leading indicators of failure/condition indicators as feature data that forms condition‐based data (CBD) signatures; such signatures can be normalized and converted into dimensionless ratios called fault‐to‐failure progression (FFP) signatures. FFP signatures form families ...