带状线电路中十字交越不连续性的具有知识神经元的神经网络模型

Neural Network with Knowledge-Based Neurons for the Modeling of Crossover Discontinuities in Stripline Circuits

本文采用一种新型的三层神经网络 ,即隐蔽层具有知识神经元的神经网络 (NNKBN)模拟带状线电路中的十字交越不连续性。NNKBN模型由一组全波时域有限差分法产生的仿真数据进行训练而建立。数值实验表明NNKBN模型在许多方面优于传统的多层感知器模型。

A novel three-layer neural network with knowledge-based neurons (NNKBN) in hidden layer has been applied to model the crossover discontinuities in stripline circuits. The NNKBN model is electromagnetically developed with a set of data that are produced by the full-wave finite-difference time-domain method. Numerical experiments show that the NNKBN model has many advantages over the conventional multi-layer perceptron model.