Abstract
Estimation of structural damage and quantification of structural integrity are critical for safe and reliable operation of human-engineered complex systems. Fatigue damage is one of the most commonly encountered sources of structural degradation in mechanical systems. Detection of incipient fatigue damage is essential for averting widespread crack growth that leads to catastrophic failures. This chapter presents online in situ monitoring of fatigue damage using the ultrasonic sensing technique that is sensitive to small microstructural changes, robust to measurement noise, and also suitable for real-time applications. A recently reported information-theoretic method of data-driven pattern recognition, called Symbolic Dynamic Filtering (SDF), has been used for real-time analysis of ultrasonic data, where the time series data in the fast scale of process dynamics are analyzed at discrete epochs in the slow scale of fatigue damage evolution. SDF includes preprocessing of ultrasonic data using wavelet transform, which is well suited for timefrequency analysis of non-stationary signals and enables noise attenuation in raw data. The wavelet-transformed data is partitioned using the maximum entropy principle to generate symbol sequences, such that the regions of data space with more information are partitioned finer and those with sparse information are partitioned coarser. Subsequently, statistical patterns of evolving damage are identified from these sequences by construction of a (probabilistic) finite-state machine that captures the dynamical system behavior by information compression. A computer-controlled fatigue test apparatus, equipped with ultrasonic sensors and an optical microscope, has been used to experimentally validate the concept of ultrasonic based real-time monitoring of fatigue damage in polycrystalline alloys. The task of fatigue damage monitoring is formulated as: (i) forward problem of pattern recognition for (offline) characterization of the statistical behavior of fatigue damage evolution and (ii) inverse problem of pattern identification for (online) estimation of the remaining useful life based on the real time ultrasonic data and the statistical information generated offline.
| Original language | English (US) |
|---|---|
| Title of host publication | Fatigue Crack Growth |
| Subtitle of host publication | Mechanics, Behavior and Prediction |
| Publisher | Nova Science Publishers, Inc. |
| Pages | 3-48 |
| Number of pages | 46 |
| ISBN (Print) | 9781606924761 |
| State | Published - Jan 2011 |
All Science Journal Classification (ASJC) codes
- General Physics and Astronomy