Heuristic-based approach: Degradation signatures and CBD signatures

This paper describes a heuristic-based approach for transforming conditioned-based data (CBD) signatures into other signatures. Those signatures are highly correlated to changes in value (dP) of a parameter of interest (P 0 ) that degrades to a level of damage at which a component and its assembly no longer functions within operation specifications: functional failure occurs. CBD signatures can be expressed in terms of feature data (FD) plus noise where the magnitude of FD changes to create a signature in response to degradation. That FD-based signature can be expressed as ratio of a current measurement (FD i ) to a nominal value (FD 0 ) to form a fault-to-failure progression (FFP) signature. Further, a present measurement can be expressed as a nominal value times a degradation function of a change (dP i ) in a parameter that has a nominal value P 0 in the absence of degradation. An FD-based signature is transformed into a fault-to-failure progression (FFP) signature by dividing a present FD by the nominal FD value and solving in terms of (dP i /P 0 ) to produce a degradation-progression signature {DPS i } which is a linearized version of an FFP signature. A DPS signature is further transformable into a functional failure signature, {FFS}, by dividing a DPS signature by a defined failure level. An FFS is amenable to processing as input to prediction algorithms because (1) its characteristic curve approaches an ideal straight-line transfer curve as noise is ameliorated and/or mitigated; (2) its data has zero or negative values in the absence of degradation; (3) its data has positive values below 100 when there is degradation below a defined level of functional failure; and (4) its data has values at or above 100 when the level of degradation is at or above a level defined as functional failure. Even in the presence of noise and other effects, such as feedback, and even when the rate of degradation is nonlinear, an FSS is still has a very close to linear transfer curve. Seven sets of models have been developed to use for transformation of CBD signatures into FFS data. The models were developed based on the characteristic shapes of degradation signatures and confirmed by physics-of-failure (PoF) analysis and failure-mode effects analysis (FMEA). This paper presents a methodology for selecting and using those models to transform CBD-based signature data into FFS data for use as input to a prediction framework of a PHM system.