Multiphysics Optimization for Tunable Filters

This article reviews efficient modeling and optimization strategies for microwave tunable filters with multiple tuning states. We first investigate a surrogate-assisted simultaneous optimization framework for multistate tunable filters, where multiple sub-surrogate models are established for different tuning states while sharing a common set of nontunable parameters. In this way, multiple tuning states can be optimized jointly within a unified framework, thereby enhancing the coordination among the design requirements of different states. Building on this, we further discuss a multiphysics optimization method based on space mapping, which combines a shared coarse model with mappings that depend on the tuning state. By exploiting low-cost electromagnetic responses as prior knowledge and using multiphysics data for corrective learning, this method improves optimization efficiency and reduces the computational cost of multiphysics design. The effectiveness of the two optimization methods is verified through a representative tunable four-pole waveguide filter example.