Research
In a general flow of structural health monitoring (SHM) and damage prognosis (DP), decisions are corrupted by uncertainty from various sources. There are different uncertainty quantification (UQ) models established in my research project, in which the uncertainty of damage features are characterized by the analytically derived probability density functions (PDFs).
Adopting the probabilistic UQ models, detection of damage occurrence is conducted via hypothesis testing, and all the decisions are made in the statistically significant sense. Thereby, the rate of true detections and false alarms are quantified through the characterized distributions, and the best detectability, i.e. the optimal trade-off between true positive and false positive, is suggested via receiver operating characteristics.
|
|
