Model-based statistical guided wave damage detection for an aluminum plate

Accurate decision-making requires understanding the factors that influence those decisions. In guided wave structural health monitoring, the first aim is to detect the presence of damage. This decision is based on the assumption that it is possible to discriminate between undamaged and damaged states. Sensing systems collect data to construct damage statistics, that is, numerical representations of how the state of the structure has changed over time. These damage statistics are designed to be sensitive to damage signatures in the data. However, environmental and operating conditions, such as the presence of temperature variations, perturb damage statistics and weaken decision-making. Temperature compensation strategies applied to reduce these perturbations are imperfect. Qualifying the performance of a temperature compensation strategy to reduce or eliminate the effects of temperature is dependent on the resulting damage statistics. Quality temperature compensation strategies must be able to recover damage statistics and increase damage detection accuracy. This article establishes a framework for evaluating the performance of damage detection methods that use temperature compensation to reduce the effects of environmental and operating conditions. Here, the scale transform and the dynamic time warping method are examined for data with varying temperature and signal-to-noise ratios. Monte Carlo simulations are used to construct probability densities and perform statistical analysis on the resulting damage statistics.