Investigation of operational and environmental variability effects on damage detection algorithms in a woven composite plate

Variability in structural health monitoring systems can result in reduced reliability by increasing the likelihood of false-positive/-negative indications of damage. It is important to understand how sources of operational, environmental, and even computational variability influence damage indicator functions. A sensitivity model-based technique, which focuses on the physical rather than the statistical nature of variability, is described. Each source of static/dynamic variability is sequentially isolated with a variability test matrix for a woven composite plate. Computational sources of variability are investigated by comparing two different damage detection algorithms (i.e., transmissibility and embedded sensitivity). It is determined that by using a piecewise variability feedback process, certain parameters of the frequency response measurement and analysis (e.g., frequency band, input–output locations, etc.) can be chosen to reduce sensitivity of the damage indices to variability. It is also shown that the sensitivity due to changes in the sensor frequency bandwidth accounts for the largest source of static variability. Finally, by using a root mean square normalization procedure, static and dynamic sources of variability can be compared with changes in damage indicators due to actual damage. It is shown that damage detection algorithms can be improved by selecting specific frequency ranges that accentuate damage indicators while minimizing the effects of variability.