Using NU-SSGA to reduce the searching time in inverse problem of a buried metallic object

We describe an inverse scattering problem with the aim of reducing the computation time for recovering the details of a perfectly conducting cylindrical object buried in a half-space. First, we use Fourier-series or cubic-spline methods to describe the shape and reformulate the inverse problem into an optimization one. Then we solved it by the improved steady-state genetic algorithm (SSGA) and simple genetic algorithm (SGA) respectively and compare the cost time in finding out the global extreme solution of the objective function. It is found the searching ability of SSGA is much powerful than that of the SGA. Even when the initial guess is far away from the exact one, the cost time for converging to a global extreme solution using by SSGA is much less than that by SGA. Numerical results are given to show that the inverse problem by using SSGA is much better than SGA in time costing.