Statistical pattern recognition on fatigue damage detection and monitoring

• Main Author: Lim, Peiyi.
• Other Authors: Soh Chee Kiong
• Format: Final Year Project
• Language: English
• Published: 2013
• Subjects: DRNTU::Engineering::Civil engineering
• Online Access: http://hdl.handle.net/10356/53833
• Institution: Nanyang Technological University

Fatigue is a type of localized material failure resulted from cyclic loadings. Fatigue failure begins with the initiation of a crack and propagates progressively under repeated loading and unloading. When the crack size reaches a certain stage, the structure faces rapid fracture which can have catastrophic effects such as loss of lives and/or properties. Signs of fatigue cracks are usually inconspicuous especially when cracks are initiated within the structure. It takes time for the crack to grow to the surface for it to be detected, hence its hazardous nature. As technology advances, damage detection and monitoring become even more imperative in order to minimize the occurrence of evitable catastrophic events, especially so in the field of Structural Health Monitoring (SHM). The ability to predict the residual useful life of a structural system becomes an added advantage as it provides a comprehensive awareness on the structural health state. This research aims to propose an effective Statistical Pattern Recognition (SPR) approach in fatigue damage detection and monitoring through experimental investigation. The report also comprises of the review on relevant studies done by various researchers using different SPR methodologies in identifying and monitoring damage. Experiments were conducted on lab-sized metallic specimen through the use of Lamb wave technique using piezoelectric transducers. A correlation factor is proposed to estimate the current crack length from the computed damage index. Last but not least, an SPR method based on an outlier analysis is proposed to detect and monitor fatigue damage. Generally, the proposed method provides a good fit of data in terms of damage detection and monitoring. However, further improvement measures and recommendations can be implemented to derive a comprehensive model, such that it can be applied to wider area and thereby provide more constructive functions in the field of SHM.