Two-stage Cognition-driven Multiphysics Optimization for Microwave Filters Design

To better design and optimize microwave filters, it is important to consider the operation of multiple physics domains, e.g., temperature, humidity, and pressure in the real-world environment. Multiphysics optimization is time-consuming because it involves multiple physical domains and often requires dealing with repetitive evaluations by sweeping different design parameters. Recently, an advanced technique exploiting artificial neural network (ANN) has shown great potential in multiphysics optimization, which improves the efficiency of the multiphysics optimization. In this work, the proposed two-stage cognition-driven technique improves the ANN method to further accelerate the multiphysics optimization. A four-pole waveguide filter is used to verify the effectiveness of the proposed technique.