Acoustic emission signature analysis of failure mechanisms in fiber reinforced plastic structures
The objective of the research program was to develop reliable pattern recognition and neural network analysis methods to determine the failure mechanism signatures in fiber reinforced plastic structures from acoustic emission (AE) data. The AE database was collected from a range of test specimens. Visual inspection and observation with a scanning electron microscope were performed to identify failure mechanisms in the specimens at various load levels. It was found that different types of specimen and structural loading yielded different types of failure. The failure mechanisms of interest were matrix cracking, debonding, delamination, and fiber breakage. Two method of analysis were used to determine the AE signatures. The first was visual AE pattern recognition. This analysis used a comparison of dissimilarities among AE correlation plots of data from different specimens. The results showed several AE signatures. The analysis also explains the correlation of material properties to failure mechanism evolution. The second analysis method was the use of neural networks to perform AE pattern recognition. The neural networks were trained using AE data in order to perform two tasks: determine the failure mechanisms and to assess the damage severity. The performance of the networks was found to be excellent for the first task and promising for the second task. The neural network was also applied to additional AE data from full-scale and coupon tests. By comparing the results from the network with visually observed damage, the network results are shown to be very reliable in determining failure mechanisms.