LSTM-Based Method for Characteristic Mode Tracking

Characteristic mode (CM) theory has been widely used for antenna design and scattering problems. However, it is always challenging to correctly track the eigenmodes for complicated structures over a wide frequency band. Due to the typical time-series features of the mode data, a method based on long short-term memory neural network (LSTM) is proposed for mode tracking. Different modes at each frequency are distinguished and classified by learning the internal features between the mode data, which can overcome the mode degradation problem and thus achieve good tracking performance. The proposed method is validated by a U-slot patch antenna.