Statistical Characterization of Damage Propagation Properties in Structural Health Monitoring

摘要: Structural health monitoring provides sensor data that follow fatigue induced damage growth in service. This information may turn be used to improve the characterization of material properties govern propagation for structure being monitored. These are often widely distributed between nominally identical structures because differences manufacturing and aging. The improved accuracy characteristics allows more accurate prediction remaining useful life (RUL) structural component. In this paper, a probabilistic approach using Bayesian statistics is employed progressively narrow uncertainty parameters spite variability error measurements. Starting from an initial, wide distribution obtained coupon tests, narrowed consecutive Detailed discussions on how construct likelihood function under given update presented. applied crack fuselage panels due cycles pressurization depressurization. It shown proposed method rapidly converges when initial size 20mm 1mm. observed narrows down grows fast, while slows slowly. property works favor will dangerous. Using identified parameters, RUL predicted with 95% confidence order obtain conservative prediction. have potential turning aircraft into flying laboratories.