Reconfigurable bandpass filter with constant absolute bandwidth based on neural network

In this paper, a novel design method for reconfigurable bandpass filter (BPF) with constant absolute bandwidth (ABW) based on neural network is proposed. A lumped-element circuit is used to design the reconfigurable BPF, which consists five series LC tanks. Three series tanks can provide three transmission poles, while two parallel tanks can introduce two same transmission zeros (TZs). This method allows designing BPFs with tunable center frequency and constant ABW. A set of inductance and capacitance values for the center frequency and ABW requirements are optimized by using neural network. Then, optimize the capacitance values to adjust the center frequency with fixed inductance values by the three series tanks and the TZ by the parallel tanks. In order to verify this design method, three varactor diodes are employed to achieve the reconfigurable BPFs with the constant ABW under different work states. Experiments show that this method can optimize multiple LC circuit parameters of BPFs, which meet design requirements, with the same ABW and different center frequencies effectively and accurately