Behavioral pattern identification for structural health monitoring in complex systems

摘要: Estimation of structural damage and quantification integrity are critical for safe reliable operation human-engineered complex systems, such as electromechanical, thermofluid, petrochemical systems. Damage due to fatigue crack is one the most commonly encountered sources degradation in mechanical Early detection essential because resulting could potentially cause catastrophic failures, leading loss expensive equipment human life. Therefore, enhanced availability, it necessary develop capabilities prognosis estimation impending onset wide-spread structures. This dissertation presents information-based online sensing using analytical tools symbolic time series analysis ( STSA). Anomaly STSA a pattern recognition method that has been recently developed based upon fixed-structure, fixed-order Markov chain. The procedure built principles Symbolic Dynamics, Information Theory Statistical Pattern Recognition. demonstrates real-time monitoring on data ultrasonic signals. changes measured monitor evolution damage. Real-time anomaly presented solution forward (analysis) problem inverse (synthesis) problem. (1)  - primary objective identification statistical signals gradual (2) the infer anomalies from observed real information generated during A computer-controlled special-purpose test apparatus, equipped with multiple devices (e.g., ultrasonics optical microscope) analysis, used experimentally validate early anomalous behavior. sensor integrated software module consisting algorithm Experiments have conducted under different loading conditions specimens constructed ductile aluminium alloy 7075 T6. The also investigated application other engineering disciplines. Two applications include combustion instability generic thermal pulse combustor model whirling phenomenon typical misaligned shaft.