Antenna Modeling Based on Image-CNN-LSTM
Zhiwei Zhu, Yubo Tian, Jinlong Sun
Abstract
In response to the time-consuming and computationally intensive issues associated with using full-wave electromagnetic simulation software exploiting global optimization method for antenna performance analysis, an efficient deep learning-based strategy is proposed and applied to rapid modeling of antennas. Considering the outstanding performance of Convolutional Neural Network (CNN) in pattern recognition and the efficiency of Recurrent Neural Network (RNN) with Long Short-Term Memory (LSTM) structures in processing sequential data, a hybrid CNN-LSTM network model is formed by combining CNN and LSTM structures. Furthermore, to enhance the model's performance, we apply image model concept to the CNN-LSTM network, proposing an Image-CNN-LSTM hybrid network. The experimental result demonstrates that the proposed network shows significant advantages in terms of prediction accuracy and fitting effects, achieving improvement in accuracy about 40% and the coefficient of determination about 4.5% compared to the CNN-LSTM network.