Executive Summary : | RCS computation is a complex problem. It depends upon object shape, frequency, computational resources, and efficient algorithm. Efficient RCS computation is of paramount importance to both defense and civilian industries. Physical Optics-based methods and Full-Wave solvers are popular choices for RCS computation. However, an accurate computation requires time-domain/frequency-domain full-wave solvers. For complex objects, these solvers demand huge computational time and resources. But they provide scattered fields for various orientations which is not possible by using Physical Optics based computation. An accurate assessment of RCS is essential for rapid prototyping. The use of various Radar Absorbing Materials has increased the challenge of RCS computation using a full wave solver manyfold. Again, RCS optimization is also a challenging problem, given the computational load taken by each iteration. This work proposes the use of the Space mapping technique to mitigate these problems. The space mapping technique is an established technique in microwave circuit optimization. It maps the fine structure to a surrogate model through a mapping function. The surrogate model takes less time for computation, but solution are accurate as that of the fine model when mapped through the mapping function. The incident plane wave on a dielectric object takes more time than compared to its equivalent metal object, therefore its simulation time is very less. However, their RCS behavior is quite different. Therefore, it is suggested that the RCS of a metal object can be mapped to the RCS of a dielectric object through a suitable mapping function. In this way, the simulation and optimization time can be reduced considerably. This work will consider using an artificial neural network as the mapping function given their high generalization capability. |