An artificial neural network-based harmonic beamforming for time-modulated array

A harmonic beamforming technology based on artificial neural network (ANN) for time-modulated array (TMA) is proposed. By establishing an ANN composed of encoder and decoder in series, timing sequence in TMA can be optimized. In ANN, the encoder takes the pattern constraint value of the target angle as the input, and outputs the corresponding excitation value through online training. While the decoder can output the real time radiation pattern through pre-training. By utilizing the optimized excitation after training, the switch-on duration and starting time of different array elements can be obtained. Both 8-element and 16-element TMA harmonic beamforming examples with different pointing directions are provided, simulation results show that the proposed method can effectively suppress sidelobe level (SLL), control the pattern fast and accurately, and realize high directional beamforming at target angle. With the superior characteristics of short time-consuming, strong robustness and easy adjustment, it is expected to be applied in fast target search and tracking.