In this article, we propose a novel neural network of calibrated coarse model, which can obtain the optimal filter response with as little training data as possible to synthesize the entire substrate integrated waveguide (SIW) filter. By incorporating the knowledge of filter decomposition with the inverse neural network, we build a coarse model that can synthesize the dimensions of a SIW filter. However, the SIW structures are subject to a potential leakage problem due to the periodic gaps, the results of the coarse model are very different from the ideal response. We propose a novel calibrated neural network from the perspective of the coupling matrix to correct the errors generated in the coarse model. In addition, this article also proposes an equivalent de‐embedding technique, which is simpler than the thru‐reflect‐line calibration technique to accurately extract the scattering parameters of the SIW discontinuities. An H‐plane fifth order SIW filter is synthesized by the proposed model. The result shows that the SIW filter that is very close to the ideal response can be synthesized with only a few hundred training data.

中文翻译：

---

校准粗模型神经网络及其在基片集成波导滤波器设计中的应用

在本文中，我们提出了一种经过校准的粗模型神经网络，该网络可以通过尽可能少的训练数据来获得最佳滤波器响应，从而合成整个衬底集成波导（SIW）滤波器。通过将滤波器分解的知识与逆神经网络相结合，我们建立了一个可以合成SIW滤波器尺寸的粗略模型。但是，由于周期性的间隙，SIW结构会遭受潜在的泄漏问题，粗略模型的结果与理想响应有很大不同。我们从耦合矩阵的角度提出了一种新颖的校准神经网络，以纠正在粗模型中产生的误差。此外，本文还提出了一种等效的去嵌入技术，它比通过反射线校准技术更简单，可以准确地提取SIW不连续点的散射参数。提出的模型合成了H平面五阶SIW滤波器。结果表明，仅用几百个训练数据就可以合成非常接近理想响应的SIW滤波器。