Genetic algorithms are used to predict the size and location of an elliptical cavity within a solid half-space. The scattering of ultrasonic waves in the solid half-space with a cavity is modeled using Distributed Point Source Method (DPSM). DPSM which is a semi-analytical technique that utilizes Green’s function is used for modeling because this technique is more efficient than popular but not so efficient Finite Element Method (FEM). FEM is very inefficient for modeling ultrasonic wave propagation problems at high frequencies and for solving an inverse problem one needs an algorithm that can solve the forward problem efficiently. The inverse problem is solved by applying a genetic algorithm to the forward problem to determine the optimum solution. The optimum population size and number of generations are determined. Results and analysis are performed for 3 cases of unknown variables.

Ultrasonic transducer modeling is important and fundamental research for nondestructive testing of materials. Traditionally, in most nondestructive evaluation applications, the ultrasonic transducers are modeled as point sources generating spherical wave fronts, line sources generating cylindrical wave fronts, or planar surfaces generating plane wave fronts. In reality, the transducer front face has finite dimensions; it is neither point source nor planar source. This study shows how the ultrasonic field in the neighborhood of a transducer with finite dimension varies in homogeneous and nonhomogeneous media. In this thesis, the pressure variation in front of a transducer face is computed for the following three situations: (1) the transducer is immersed in a homogeneous fluid; (2) the transducer is near a fluid-fluid interface; and (3) the transducer is near a fluid-solid interface using Distributed Point Source Method For the nonhomogeneous fluid medium, both normal incidence and oblique incidence cases have been studied.