Recent progress on computational inverse techniques in non-destructive evaluation

Abstract

This chapter introduces computational inverse techniques used in nondestructive evaluation for material and structural systems. It emphasizes on several robust and practical inverse algorithms developed through the combination of different types of optimization methods, such as gradient-based methods with genetic algorithms and intergeneration projection genetic algorithms. Methods of the progressive neural networks are also discussed. A regularization technique that works particularly well with engineering illposed inverse problems is described. The efficiency and features of all these computational inverse techniques have been demonstrated to solve practical complex nondestructive evaluation problems of crack detection, force function reconstruction, and material property identification. The inverse problems are formulated into parameter-identification problems, in which a set of parameters corresponding to the characteristics can be found by minimizing error functions formulated using the measured dynamic behaviors of structures and that computed using forward solvers based on projected candidates of parameters.