COMPUTATIONALLY EFFICIENT TWO-OBJECTIVE OPTIMIZATION OF COMPACT MICROWAVE COUPLERS THROUGH CORRECTED DOMAIN PATCHING

Abstract

Finding an acceptable compromise between various objectives is a necessity in the design of contemporary microwave components and circuits. A primary reason is that most objectives are at least partially conflicting. For compact microwave structures, the design trade-offs are normally related to the circuit size and its electrical performance. In order to obtain comprehensive information about the best possible tradeoffs, multi-objective optimization is necessary that leads to identifying a Pareto set. Here, a framework for fast multi-objective design of compact micro-strip couplers is discussed. We use a sequential domain patching (SDP) algorithm for numerically efficient handling of the structure bandwidth and the footprint area. Low cost of the process is ensured by executing SDP at the low-fidelity model level. Due to its biobjective implementation, SDP cannot control the power split error of the coupler, the value of which may become unacceptably high along the initial Pareto set. Here, we propose a procedure for correction of the S-parameters’ characteristics of Pareto designs. The method exploits gradients of power split and bandwidth estimated using finite differentiation at the patch centres. The gradient data are used to correct the power split ratio while leaving the operational bandwidth of the structure at hand intact. The correction does not affect the computational cost of the design process because perturbations are pre-generated by SDP. The final Pareto set is obtained upon refining the corrected designs to the high-fidelity EM model level. The proposed technique is demonstrated using two compact microstrip rat-race couplers. Experimental validation is also provided.