Beam-forming optimization of antenna array with radome using multi-layer perceptron network

This paper investigates beamforming optimization of antenna array with radome through the use of multi-layer perceptron (MLP) network. The MLP network is found to be more effective than the projection matrix method, as it can correct beam distortion caused by the radome structure and reduce the side lobe level (SLL) by approximately 5 dB. To reduce the complexity of antenna array with radome and decrease the engineering realization difficulty of the feed network, this paper proposes a subarray partition technique. The results indicate that this technique retains some of the optimization properties of the MLP network and has significant potential for antenna array-radome designs.