摘要
导体传输线的耦合截面(CCS)是评估场线耦合效应的重要参数,CCS一直没有快速且精度较高的获取方法。针对CCS的快速获取问题,在全波分析法下用蒙特卡洛均匀抽样法获取了9000组CCS数据,将数据归一化处理后输入神经网络进行训练,建立了一个以导体传输线线长、线高、线径、线间距、入射波极化角、入射波频率为输入,CCS为输出的BP神经网络模型。选取1600组未参与训练的输入数据用该模型进行预测,将预测结果与全波分析对比,88.8%的预测数据相对误差在10%以内,且相对误差较大的数据其CCS往往较小。
The coupling cross section(CCS)of a conductor transmission line is an important parameter to evaluate the cou⁃pling effect of field lines.There has been no fast and accurate method to obtain the CCS.Aiming at the problem of fast acquisition of the CCS,9000 sets of coupling section data were obtained using the Monte Carlo uniform sampling method under the full⁃wave analysis,and the data were normalized and input to the neural network for training.A BP neural network model was established with conductor line length,line height,line diameter,line spacing,incident wave polarization angle,incident wave frequency of transmission line as the input,and the CCS as the output.1600 sets of input data that did not participate in the training were se⁃lected to predict with the model,and the prediction results were compared with the full⁃wave analysis.The results showed that 88.8%of the predicted data had a relative error within 10%,and the data with a large relative error often had a smaller CCS.
作者
周世华
赵翔
Zhou Shihua;Zhao Xiang(College of Electronic and Information Engineering,Sichuan University,Chengdu 610065)
出处
《现代计算机》
2023年第3期85-90,共6页
Modern Computer
关键词
导体传输线
耦合截面
全波分析
蒙特卡洛抽样
BP神经网络
conductor transmission line
coupling cross section
full⁃wave analysis
Monte Carlo sampling
BP neural network
