ABSTRACT In this paper, a multi-parametric analysis modelling for an ultra-wideband (UWB) antenna using artificial neural networks is proposed to predict both the UWB impedance bandwidth and multi-band notch frequencies. Two different neural models are implemented and tested using eight different algorithms. Neural model with two hidden layers having testing mean square error as 27.51 kHz proved to be an efficient one. The effectiveness of the optimised models is further validated using an additional dataset which has not been used during training and/or testing phases. A prototype is fabricated and characterised. Close agreement between simulated and experimental results is validated for the authenticity of the suggested work.

中文翻译：

---

超宽带天线干扰抑制的多参数人工神经网络建模

摘要 在本文中，提出了一种使用人工神经网络的超宽带 (UWB) 天线的多参数分析建模，以预测 UWB 阻抗带宽和多频带陷波频率。使用八种不同的算法实现和测试了两种不同的神经模型。测试均方误差为 27.51 kHz 的具有两个隐藏层的神经模型被证明是一种有效的模型。使用在训练和/或测试阶段未使用的附加数据集进一步验证优化模型的有效性。制造和表征原型。模拟和实验结果之间的密切一致性验证了建议工作的真实性。