A New Interval Finite Element Approach for Structural System Identification and Damage Detection under Uncertainty

Georgia Tech Research Corporation, Atlanta GA

Investigators

Abstract

Identifying material properties of existing structural systems is crucial for monitoring their performance and ensuring their safety. Knowledge about these properties allows the determination of the magnitudes and the locations of any deterioration occurring in the system. Such knowledge will allow the proper required actions to avoid any failures. The identification procedures require a significant amount of measured data that typically are associated with uncertainty. Accounting for uncertainty due to different factors is one of the most challenging tasks in structural system identification and damage detection. In this project, uncertainty in measurements due to device tolerances and field conditions will be treated in terms of possible values expressed as intervals, without any prior assumptions on their nature. The rigorous consideration of uncertainty in system identification will lead to a more reliable and dependable predictions of their behavior. Such reliable and dependable predictions are crucial for the safety of civil infrastructure systems, and will contribute to the reduction of life endangering failures and disasters. The research team will offer research opportunities to high school students through summer camps. While interval-based methods and system identification are both very powerful and well-developed techniques within their respective fields, thus far they have scarcely been used together in the context of inverse problems under uncertainty. By investigating the best ways of analytically, numerically and computationally handling these two methodologies in conjunction with each other, we shall develop a powerful class of techniques not only for the problem at hand, but for a vast number of other inverse problems with partial differential equation (PDE) constraints. In this research an emphasis will be placed on the use of Interval Finite Element Methods (IFEM) in the context of optimization to solve for inverse problems that arise in structural system identification and damage detection under uncertainty without prior assumptions on its nature. This is therefore expected to be the first step of a long lasting research program which will be both broad in its scope of applications across many fields and at the same time remain coherently focused in its underlying computational and mathematical methodologies.