The Damage Detection module consists of two different damage detection algorithms that have been developed for long-term monitoring of structures. The algorithms are robust to the natural change of a structure’s dynamic behavior due to the changes of ambient conditions and can be set up to not reacting on such changes. The plugin also includes a Control Chart that, as input, can take a user-definable set of damage indicators and unify them to a single control value with corresponding statistical based threshold. This control chart can automatically notify (alert) when the threshold is being passed after an analysis. The notification can be sounds, emails or web service calls to third party systems.
The features are summarized below:
Classic and Robust Damage Detection.
Automatic damage indicator updates when new measurements are uploaded.
Reference State Validation for visual validation of the null space estimation of the reference state model.
Hotelling Control Chart unifying damage indicators into a single indicator with a single statistical based threshold.
In the figure below, the damage detection is demonstrated using data from a bridge in Italy, that was measured for a period before damaging it. There are eight reference measurements that are presented by the first eight green vertical bars in the diagram.
The level of these bars shows the level of the damage indicators under normal conditions. From the first six measurements, a baseline (reference) model is determined by the module, and a threshold is automatically estimated based on the statistical evaluation of the damage indices of the six reference measurements.
As seen, the last two reference measurements stay under the threshold which indicates that the bridge is still undamaged. The remaining 14 measurements are all acquired after the bridge experienced damage, and the associated damage indicators all pass the threshold significantly and indicate a permanent damage.