ANN-enabled Mapping Between Equivalent Circuit Model and Physical Field Model for Tunable Bandpass Filter

In this paper, an ANN model development approach is outlined and studied for the development of tunable bandpass filter (tBPF). Circuit models are fast and efficient for design and optimization while physical models based on electromagnetic full-wave fields are accurate but computationally expensive and slow. In this work, an equivalent circuit model is developed from a full-wave tBPF structure. Then, equivalent circuit parameters are generated and mapped to tunable geometric parameters of the physical field model of the tBPF. The optimization is thus carried out on the equivalent circuit model. Finally, the geometric parameters are obtained by the ANN model from the optimized equivalent circuit parameters. A practical example of tBPF is used to demonstrate the proof of concept.