Damage detection index based on statistical inference and PCA
Author
Mujica, L.E.; Ruiz, M.; Pozo, F.; Rodellar, J.
Type of activity
Presentation of work at congresses
Name of edition
8th International Workshop on Structural Health Monitoring
Date of publication
2011
Presentation's date
2011-09-12
Book of congress proceedings
Structural health monitoring 2011: condition-based maintenance and intelligent structures : proceedings of the 8th International workshop on structural health monitoring, Stanford University, Stanford, CA, September 13-15, 2011
This paper is focused on the development of new estimators propounding if someone statistical law could estimate or infer a system without damage knowing its reliability. This new measurement considers each experiment, and consequently, each
projection to the PCA model as a random variable. An in-depth statistical analysis is
performed for SHM. PCA projections are obtained from the undamaged structure
(baseline projection). If these projections are considered as the set of possible results
(popu...
This paper is focused on the development of new estimators propounding if someone statistical law could estimate or infer a system without damage knowing its reliability. This new measurement considers each experiment, and consequently, each
projection to the PCA model as a random variable. An in-depth statistical analysis is
performed for SHM. PCA projections are obtained from the undamaged structure
(baseline projection). If these projections are considered as the set of possible results
(population), then the new projections from the current structure (healthy or not) are
defined as random samples. Therefore, the probability distribution of the baseline
projection can be found. This new distribution can make an inference about the state of the structure and determine if there is damage in it. Consequently, the relative likelihood of each new projection is determined. If the new projection is strongly related with the population, then the structure is healthy. Otherwise, the relation indicates the damage.
This paper is focused on the development of new estimators propounding if someone statistical law could estimate or infer a system without damage knowing its reliability. This new measurement considers each experiment, and consequently, each projection to the PCA model as a random variable. An in-depth statistical analysis is performed for SHM. PCA projections are obtained from the undamaged structure (baseline projection). If these projections are considered as the set of possible results (population), then the new projections from the current structure (healthy or not) are defined as random samples. Therefore, the probability distribution of the baseline projection can be found. This new distribution can make an inference about the state of the structure and determine if there is damage in it. Consequently, the relative likelihood of each new projection is determined. If the new projection is strongly related with the population, then the structure is healthy. Otherwise, the relation indicates the damage.
Citation
Mujica, L.E. [et al.]. Damage detection index based on statistical inference and PCA. A: International Workshop on Structural Health Monitoring. "Structural health monitoring 2011: condition-based maintenance and intelligent structures : proceedings of the 8th International workshop on structural health monitoring, Stanford University, Stanford, CA, September 13-15, 2011". Stanford, CA: Destech, 2011.