The goal of the research is to develop new computational techniques for nondestructive evaluation (NDE) of structural components that will provide fast, reliable, and highly accurate quantitative evaluation of the current state of structural properties. Algorithms will combine efficient computational models of the structural response with optimization approaches to identify the structure's state of degradation with a given set sensor measurements. To obtain high level of efficiency and accuracy, novel model reduction techniques will be established that are tailored for NDE applications. In addition, the algorithms will define the optimal methods of exciting the structure and measuring the structural response to provide a high-resolution local description of the damage. The testing methods will be designed to maximize the sensitivity of the structural properties to the test parameters and will be applicable to a host of NDE applications from nuclear reactor to bridge structures.
The research results will enable engineers to more efficiently and accurately predict the current state of degradation and the remaining life of a structure, leading to more timely response to hazardous conditions and fewer false alarms and closures. The algorithms developed in the project will be freely disseminated to the community for wide-spread use and future developments. The research will provide advanced training to graduate and undergraduate students and establish a summer research program for undergraduates and pre-college students in collaboration with the office of diversity to engage students from traditionally underrepresented groups in engineering.
The objective of this project was to advance computational methods for solving inverse problems related to nondestructive evaluation of structural materials. In particular this work focused on strategies to design testing methods for optimal nondestructive evaluation accuracy and robustness. The major outcome of the work in nondestructive test design was the creation and testing of a method to optimally utilize the resources (e.g., sensors and actuators) in the implementation of nondestructive testing protocols. A design procedure was created to ensure that the resulting nondestructive test produces the most relevant and non-redundant information relating to the state of the system being evaluated, and this procedure was tested and confirmed to provide substantial benefits in terms of accurate system evaluation with the designed tests. In addition, the final design procedure also included a means to ensure that the nondestructive test designed would be insensitive to system variables that may have some amount of unknown variation, and therefore improve robustness of the resulting test methods as well. The project also investigated procedures improve the accuracy and efficiency (in terms of the time to produce an estimate) of computational nondestructive evaluation methods to utilize such nondestructive testing data. An optimization process was investigated to solve nondestructive evaluation problems, and it was determined that a simple change in the objective of the optimization process substantially improves the quality of the resulting nondestructive evaluation solution estimate. More specifically, by simply breaking the measurement data provided by the nondestructive test into multiple objectives, instead of the typical approach of combining this data into one objective, and then using a multi-objective optimization technique, the evaluation procedure was shown to be substantially more accurate, efficient, and robust in the presence of measurement noise.