Robustness of moving-window Ib-value analysis for acoustic-emission-based fracture precursor detection under severe background noise

Background noise in operating nuclear power plants masks the acoustic emission (AE) signals of micro-cracks and hinders early fracture-precursor detection. A moving-window Ib-value methodology is proposed and validated by a proof-of-concept tensile test on a notched SM45C carbon steel specimen. For 651 crack signals screened by source location, the energy-based b-value (Aki's maximum likelihood method) and the Ib-value were evaluated, and detection-threshold elevation was emulated by removing up to 50% of the low-amplitude data. The cumulative approach diluted the precursor and never reached the alarm criterion of 1.0, whereas the 100-hit moving window immediately captured the shift of the crack mode from shear to tensile. Under increasing noise, low-amplitude micro-crack signals were selectively lost (survival 100% to 40%, against 100% for macro-cracks), biasing the b-value by up to −19.8%, while the Ib-value, which evaluates only the distribution's statistical center, stayed within ±4.9%. Across 0–40% hit loss it still crossed the criterion 7.4–11.8 s before rupture, at 98.4–99.3% of peak stress. The moving-window Ib-value is therefore markedly less sensitive than the b-value to the selective loss of low-amplitude AE events caused by threshold elevation. Generalization to full background-noise robustness, which also involves waveform superposition and parameter distortion, requires further validation.