Prognostics and Health Management

4Heuristic‐Based Approach to Modeling CBD Signatures

4.1 Introduction to Heuristic‐Based Modeling of Signatures

Stresses and strains in systems, cyclic or otherwise, often cause accumulating fatigue damage in prognostic targets, resulting in changes in one or more measurable signals called condition‐based data (CBD). Those changes are leading indicators of failure that can be extracted, conditioned, and transformed into dimensioned feature data (FD) and then into dimensionless fault‐to‐failure progression (FFP) signature data. Such data is highly correlated to the progression of a prognostic target from a state of 100% health (zero or insignificant degradation) to a state of zero health (degraded to a level defined as functionally failed). FFP signature data can be further conditioned and transformed into an FFP‐based functional‐failure signature (FFS) and/or into a degradation‐progression signature (DPS) and then into a DPS‐based FFS (Viswanadham and Singh 1998; Hofmeister et al. 2013, 2016, 2017; Kwon et al. 2010).

Referring to Figures 4.1 and 4.2, the development or selection of a set of models for CBD signatures of interest is a first step in transforming CBD signature data into an FFS for processing by a predication system to produce prognostic information in a prognostics and health management/monitoring (PHM) system. While it is certainly possible to use, for example, failure‐mode and effects analysis (FMEA) and physics of failure (PoF) methods and to develop a degradation‐signature ...