Prognostics and Health Management

3Failure Progression Signatures

3.1 Introduction to Failure Signatures

Chapter 2 introduced three classical prognostic approaches for prognostics and health management/monitoring (PHM): model driven, data driven, and hybrid driven. You were also introduced to usage‐based and condition‐based approaches. You learned the primary disadvantages of classical and usage‐based approaches for prognostics: they are not applicable to a specific prognostic target in a system, and/or they are nondeterministic and not suitable for application to prognostic targets, and/or it is complex to adapt them to sensor data. You also learned that leading indicators of failure can be extracted from sensor data and collected to form condition‐based data (CBD) signatures; the modeling and processing of such signatures is a condition‐based approach to condition‐based maintenance (CBM). Figure 3.1 shows the relationship of an approach using CBD signatures to classical PHM approaches; although the block diagram indicates the approaches are different, a conditioned‐based approach often employs analysis and modeling techniques such as reliability modeling, physics of failure (PoF) analysis, and failure mode and effect analysis (FMEA) (Hofmeister et al. 2013, 2016, 2017; Medjaher and Zerhouni 2013; Pecht 2008).

Figure 3.1 Diagram of classical and CBD prognostic approaches for PHM systems.

Source: based on Pecht ...