面向行波管放大器的多状态行为模型表征方法研究  

Research on multi-state behavior model characterization method for traveling wave tube amplifiers

作　　者：邹峰 常宗煜 胡欣 王刚 ZOU Feng;CHANG Zongyu;HU Xin;WANG Gang(Aerospace Information Research Institute,Chinese Academy of Sciences,Beijing 100190,China;School of Electronic,Electrical and Communication Engineering,University of Chinese Academy of Sciences,Beijing 100049,China;School of Electronic Engineering,Beijing University of Posts and Telecommunications,Beijing 100876,China)

机构地区：[1]中国科学院空天信息创新研究院,北京100190 [2]中国科学院大学电子电气与通信工程学院,北京100049 [3]北京邮电大学电子工程学院,北京100876

出　　处：《电子科技大学学报》2025年第2期197-202,共6页Journal of University of Electronic Science and Technology of China

摘　　要：近年来,以深度神经网络为代表的人工智能技术被应用于功率放大器的行为模型构建中,高精度的非线性拟合度可以满足功率放大器行为模型表征的要求,但仅适用于单一工作状态。随着对行波管放大器输入输出特性的深入研究,输出信号受到输入端激励信号的频率和温度变化等多因素的影响,如何基于深度神经网络构建面向行波管放大器的多状态行为模型亟需研究。该文提出一种面向行波管放大器的多状态神经网络建模方法,引入嵌入编码向量表征行波管放大器的多种工作状态,通过增加跳跃连接构造多状态行为模型以避免梯度消失的问题。实验结果表明,与传统方法相比,该方法能够构建表征行波管放大器的多种工作状态,且不会随着模型规模的增加而损失模型精度。In recent years,artificial intelligence technologies represented by deep neural networks have beenapplied in the construction of behavioral models for power amplifiers.High precision nonlinear fitting can meet therequirements of characterizing the behavioral models of power amplifiers,but it is still only applicable to a singleworking state.With the in-depth study of the input-output characteristics of traveling-wave tube amplifier(TWTA),the output signal is affected by multiple factors such as the frequency of the input excitation signal and temperaturechanges.Therefore,it is urgent to study how to construct a multi-state behavior model for TWTA based on deepneural networks.This article proposes a multi-state neural network modeling method for TWTA.This methodintroduces embedded encoding vectors to characterize the various working states of TWTA,and innovativelyconstructs a multi-state behavior model by adding skip connections to avoid the problem of gradient vanishing.Theexperimental results show that compared with traditional methods,this method can construct multiple operatingstates that characterize TWTA without losing model accuracy as the model size increases.

关 键 词：行为模型 行波管放大器 跳跃连接 嵌入向量 

分 类 号：TN92[电子电信—通信与信息系统]