Dynamic Beam Control Method for Multiple Moving Targets in Microwave Wireless Power Transmission

ABSTRACT This paper proposes an efficient real‐time control method for near‐field electric fields of array antennas, aimed at achieving energy focusing for multiple moving energy‐receiving targets in near‐field microwave wireless power transmission (MWPT) systems. The proposed method utilizes a Conditional Deep Convolutional Generative Adversarial Network (CDCGAN) to rapidly predict the phase excitations required for focusing. By employing geometric modeling, the network directly constructs the ideal electric field distribution on the focal plane and samples it as training data, thereby avoiding the conventional iterative optimization process and significantly improving the efficiency of training data acquisition. Furthermore, the near‐field electric field computation module is embedded into the training processes of both the generator and discriminator, while the Mean Squared Error (MSE) loss and Binary Cross‐Entropy (BCE) loss are jointly applied to balance pixel‐level accuracy and model generalization. The results show that the proposed approach can achieve rapid near‐field radiation field control—including flat‐top focusing and multi‐focus patterns—within 10 ms, while effectively suppressing sidelobes. Furthermore, the model also demonstrates excellent scalability. These results demonstrate the method's strong potential for enhancing transmission efficiency and adaptability in dynamic energy‐receiving target scenarios of MWPT systems.