On-Line Time-Series Damage State Estimation Using Correlation Analysis and Ultrasonic Broadband Active Sensing

This paper proposes an approach to estimate the fatigue-induced damage state of a metallic structure under biaxial cyclic load. An ultrasonic based technique for broadband active sensing is used. A novel damage index estimation technique based on dual sensor signals is developed that removes the contribution from high frequency input dependant noise. The scalar damage index at any particular damage condition is evaluated using a nonparametric system identification approach such as correlation analysis. The time series estimation of the damage indices (or damage states) shows good correlation with visual measurements of crack length. The time series 2σ error bound was also evaluated to study the effect of measurement noise on damage state estimation.