Open Access Open Access  Restricted Access Subscription or Fee Access

Sensitivity Analysis and Uncertainty Evaluation in the Design of Structural Health Monitoring Systems

CHRISTINE M. SCHUBERT KABBAN, BIN LIN, YEASIN BHUIYAN, CHERYL P. EDELMANN, VICTOR GIURGIUTIU

Abstract


The complexity of uncertainty evaluation for autonomous SHM systems is compounded by the numerous conditions and variables for which the SHM system must demonstrate reliable and consistent performance. Performing all such tests on real structures is often infeasible. This work examines the feasibility of using a statistical experimental design (DOE) framework in the simulation environment in order to conduct a multi-factored sensitivity study to inform subsequent studies and, eventually, the validation of the SHM system. Specifically, we examined the simultaneous effect of varying sensor placement, flaw location, frequency, and sensor size on the performance of a known piezoelectric wafer active sensor (PWAS) in a simulation environment. Using fractional experimental designs, we demonstrate the replicability of using ½ the total number of experimental runs to identify the signal feature exhibiting the most sensitivity to system design factors and the factors exhibiting the greatest effect (greatest variability) on the SHM response. These factors varied depending on the damage index and feature used for analysis. Specifically, emphasis and care should be taken when designing an SHM system based upon the correlation coefficient and difference in peak amplitude with respect to the transmitter angle and receiver angle, in addition to frequency for the difference in peak amplitude. These factors accounted for between 90-98% of all the variability observed in the SHM response for these signal features. The results of this experiment demonstrate: 1) the usefulness of designed experiments for efficient uncertainty evaluation in SHM systems, 2) the sensitivity of damage indices to varying SHM factors, and 3) the ability of simulation experiments to effectively inform subsequent material and real-time experimentation. These initial results have provided the basis for more extensive demonstrations on complex simulation experiments in addition to laboratory and active material.


DOI
10.12783/shm2017/14215

Full Text:

PDF

Refbacks

  • There are currently no refbacks.