An Efficient Gradient-Based Hybrid Parameter-Topology Optimization for Antenna Design

A gradient-based hybrid parameter-topology (HPT) method is proposed, which can optimize the structural parameter and 2-D topology simultaneously. The 2-D topology is formed by the pixel representation method, which frees topological variables from the limitation of the finite element mesh, and topological variables are independent of parametric variables. Adjoint sensitivity analysis is used to obtain the gradient information of the objective function. The calculation time and accuracy of full-wave electromagnetic (EM) simulation are closely related to the finite element mesh. Since the topology variables are independent of the mesh, a multi-fidelity model method can be applied to the optimization process, which not only ensures the optimization accuracy, but also improves optimization efficiency. Finally, three antenna optimization examples are used to demonstrate that the proposed HPT method can not only achieve fast optimization, but also produce structures with smooth boundaries. Examples show that the multi-fidelity model method can reduce the optimization time by 15%.