Sensitivity analysis of temperature effects on guided wave-based damage detection

Abstract

This paper presents a sensitivity analysis of guided wave-based damage detection to temperature uncertainty in the signal measurements. The effects of temperature variations on the damage index (DI) and consequently damage detection under varying operational temperature is investigated. Measurements from both the pristine and damage states are collected using an environmental chamber at a 5°C temperature interval. The correlation coefficient DI is utilized for the damage detection and a stochastic metamodel is developed to study the influence of temperature by implementing Gaussian Process Regression (GPR) and Monte Carlo (MC) sampling. The separation range between the pristine and damage states is used to quantify the system’s detection performance in the temperature difference range of ±20°C, imposing constrains on the acceptable false negative and false positive rates. Then, the sensitivity of the separation range is used to identify how variations on the baseline and current temperature affect the system’s reliability. The results obtained indicate that the system’s performance is improved when the baseline temperature is selected to be lower than that of the current temperature. The proposed method is validated on a second panel with a similar damage located at the corner and it is also applicable to other structures with varied requirements of temperature conditions and damage detection reliability.