Assessment of multi-objective optimization for nondestructive evaluation of damage in structural components. (June 2014)

• Record Type: Journal Article
• Authors: Wang, Mengyu; Brigham, John C
• Other Names: Smith Ralph guest-editor.; Wereley Norman M guest-editor.; Myers Oliver guest-editor.; Dragoni Eugenio guest-editor.; Ruggiero Eric guest-editor.; Loh Kenneth J guest-editor.
• Is Part Of: Journal of intelligent material systems and structures. Volume 25:Number 9(2014:Jun.)
• Page Start: 1082
• Page End: 1096
• Publication Date: 2014-06
• Subjects: Multi-objective optimization -- inverse characterization -- nondestructive evaluation -- genetic algorithm -- computational inverse mechanics; Smart materials -- Periodicals; Intelligent control systems -- Periodicals; Artificial intelligence -- Periodicals; Matériaux intelligents -- Périodiques; Commande intelligente -- Périodiques; Intelligence artificielle -- Périodiques; 620.11
• DOI: 10.1177/1045389X13494933
• Languages: English
• ISSNs: 1045-389X

Abstract

A multi-objective optimization-based computational approach to nondestructive evaluation of damage in structural components, and more generally in solid continua, is discussed and numerically evaluated. The multi-objective approach provides a substantial improvement in the capabilities to traverse the optimization search space to minimize the measurement error and produce accurate damage estimates. Through simulated test problems based on the characterization of damage in structural steel components, including internal pipe surface geometry as well as material loss within a plate structure utilizing steady-state dynamic measurements of outer surface displacement, a multi-objective genetic algorithm optimization approach is shown to provide substantial computational improvement over single-objective strategies. The multi-objective approach consistently and efficiently produces more accurate characterization results in contrast to equivalent single-objective strategies. More importantly, the multi-objective approach is shown to exhibit consistently better tolerance to test measurement noise and measurement sparsity. Moreover, the multi-objective strategy was found to provide improved diversity in the solution estimates for ill-posed problems, which is an important step leading to insight into the necessary changes to the testing or parameterization to subsequently produce more accurate and unique solutions to such inverse characterization problems.