Dynamic time warping for temperature compensation in structural health monitoring

Abstract

Guided wave structural health monitoring uses ultrasonic waves to identify changes in structures. To identify these changes, most guided wave methods require a pristine baseline measurement with which other measurements are compared. Damage signatures arise when there is a deviation between the baseline and the recorded measurement. However, temperature significantly complicates this analysis by creating misalignment between the baseline and measurements. This leads to false alarms of damage and significantly reduces the reliability of these systems. Several methods have been created to account for these temperature perturbations. Yet, most of these compensation methods fail in harsh, highly variable temperature conditions or require a prohibitive amount of prior data. In this paper, we use an algorithm known as dynamic time warping to compensate for temperature in these harsh conditions. We demonstrate that dynamic time warping is able to account for temperature variations whereas the more traditional baseline signal stretch method is unable to resolve damage under high temperature fluctuations.