( )=Δ cum( ) Δ ( ) = cum( )− cum( −1) ( )− ( −1) (4) This metric expresses the intensity of AE hit accumulation per unit stress increment. At the same time to reduce step-to-step fluctuations, ( )is smoothed using a moving average window (default window in pilot analysis was set as length of =5): ෨ ( )= 1 ( ), + = − =2 +1 (5) To limit non-representative AE activity related to specimen relaxation and early contact effects, the algorithm ignores points below a user-defined stress threshold ignore(default ignore =5MPa). The corresponding force threshold is: ignore = ignore 1000 (6) Thus only points with ≥ ignore are considered. A baseline segment of ෨ is taken from the initial portion of the accepted data (default base =40 rows). The baseline median is: med =median( ෨(1. . base)) (7) The median absolute deviation (MAD) is computed as: MAD=median(∣ ෨( ) −med ∣) =1.. base (8) A detection threshold is then defined as: Thr =med+ ⋅ MAD (9) where is a user parameter (default =5). To avoid false detection under very low AE activity, an additional constraint is applied: ( ) ≥ min (10) where minis a user parameter (default min =5 hits per step). The Kaiser point is defined as the first data step satisfying all conditions: ෨ ( )>thr ∧ ( ) ≥ ignore ∧ ( ) ≥ min (11)
RkJQdWJsaXNoZXIy MTM0Mzk2