Microwave filter tuning for different center frequencies based on Artificial Neural Network and phase compensation

This paper proposes a new approach for tuning microwave filters, which are tunable in a certain frequency range. An Artificial Neural Network (ANN) is used to build a filter inverse model, which maps scattering parameters into tuning elements deviations. Once the model is built, it may be applied for tuning filters of the same type on different center frequencies provided that the filter design allows to shift its band. Described considerations showed that a transformation of phase has to be performed on working signal. Five different types of phase compensation methods are proposed based on experiments performed on 4^th order filter. Such compensation modifies scattering parameters in a way that the filter can be tuned for different frequencies using the same inverse model. Experiments performed on 7^th order filter verified this approach.