Shape Reconstruction of Two-Dimensional Acoustic Obstacle Using Particle Swarm Optimization

An inverse scattering technique for estimating the unknown shape of a two-dimensional acoustic obstacle, based on scattered pressure field measurements, is proposed. The pressure field is measured around the obstacle domain at distinct points when the obstacle is excited by incident plane waves with different angles of incidence. The contour of the obstacle is described by a parametric model utilizing cubic B-splines, while the obstacle boundary is considered acoustically either soft or hard. Given an estimate of the obstacle shape the scattered field is computed based on the boundary element method. Then the shape is updated iteratively by minimizing a cost function representing the discrepancy between the measured and the estimated field. To minimize the cost function the particle swarm optimization algorithm is adopted, which belongs to the general class of evolutionary algorithms. In the present study, two different operation modes of the particle swarm optimization, the asynchronous and the synchronous one, are examined and applied to simulated scattered pressure field data. The performance of the algorithm with respect to different original obstacle shapes and in cases of noisy simulated measurements is investigated.