In this article an S-shaped rectangular microstrip patch antenna (RMSA) is investigated for S-band applications using artificial neural network (ANN). The authors have done the parametric study of different radiating structures to obtain S-shaped RMSA. The size of inserted notches on the radiating patch for achieving wideband operation is computed through multilayer perceptron artificial neural network (MLP-ANN) over a desired range of its performance effecting parameters such as frequency, gain, directivity, antenna efficiency, and radiation efficiency. MLP-ANN model is trained and tested with seven different algorithms. The research found that Levenberg-Marquardt (LM) training algorithm takes less computational time with better accuracy for computation of notches size on radiating patch over a priory defined performance parameters. To verify the work, a prototype of S-shaped RMSA is physically fabricated on foam substrate and tested experimentally. The experimental results are in good agreement with the simulated results that are produced with ANN.

中文翻译：

---

使用多层感知器神经网络对 S 型 RMSA 进行 S 波段应用的性能研究

在本文中，使用人工神经网络 (ANN) 研究了 S 形矩形微带贴片天线 (RMSA) 的 S 波段应用。作者对不同的辐射结构进行了参数研究，得到了 S 形的 RMSA。通过多层感知器人工神经网络 (MLP-ANN) 在其性能影响参数（例如频率、增益、方向性、天线效率和辐射效率）的所需范围内计算用于实现宽带操作的辐射贴片上插入的凹口的大小。MLP-ANN 模型使用七种不同的算法进行训练和测试。研究发现，Levenberg-Marquardt (LM) 训练算法在计算辐射贴片上的凹口尺寸时占用的计算时间更少，精度更高，超过先验定义的性能参数。为了验证这项工作，在泡沫基材上物理制造了一个 S 形 RMSA 原型并进行了实验测试。实验结果与人工神经网络产生的模拟结果非常吻合。