一种基于神经网络的纳秒脉冲波形重建方法

吕东辉; 程杰; 李锐; 张楠; 张立刚

doi:10.11884/HPLPB202537.240342

一种基于神经网络的纳秒脉冲波形重建方法

doi: 10.11884/HPLPB202537.240342

吕东辉^{1, 2,},
程杰^{1, 2, ,},
李锐^{1, 2},
张楠^{1, 2},
张立刚^{1, 2}

1.
西北核技术研究所，西安 710024
2.
先进高功率微波技术重点实验室，西安 710024

基金项目: 先进高功率微波技术重点实验室课题

详细信息

作者简介:
吕东辉，lvdonghui@nint.ac.cn

通讯作者:
程　杰，chengjie@nint.ac.cn。

中图分类号: TM836
计量
- 文章访问数: 38
- HTML全文浏览量: 21
- PDF下载量: 3
- 被引次数: 0
出版历程
- 收稿日期: 2024-09-24
- 修回日期: 2024-11-27
- 录用日期: 2024-11-27
- 网络出版日期: 2024-12-11

A nano-second pulse waveform reconstruction method based on neural network

Lü Donghui^{1, 2
,},
Cheng Jie^{1, 2
, ,},
Li Rui^{1, 2},
Zhang Nan^{1, 2},
Zhang Ligang^{1, 2}

1.
Northwest Institute of Nuclear Technology, Xi’an 710024, China
2.
Key Laboratory of Advanced Science and Technology on High Power Microwave, Xi’an 710024, China

摘要

摘要: 针对一种由高速数采通道存在寄生参数、带宽不足导致的纳秒脉冲测量波形畸变的问题，提出了一种基于神经网络的波形重建方法。通过单一神经网络辨识高速数采畸变波形与示波器参考波形的局部映射关系，通过神经网络序列完成全局波形的重建。验证实验表明所提出的方法可以明显缓解高速数采波形的边沿变缓、过冲等问题，波形功率估计精度提高32.5%，能够显著改善高速数采的频响特性。
- 神经网络 /
- 波形重建 /
- 纳秒脉冲 /
- 波形畸变 /
- 频响特性
Abstract: A new method of waveform reconstruction based on neural network is proposed to solve the problem of nano-second pulse distortion, which is caused by the existence of parasitic parameters and insufficient bandwidth in high-speed digital acquisition channels. The local mapping relationship between the distortion waveform acquired by the high-speed digital acquisition system and the reference waveform obtained from the oscilloscope is identified through single neural networks. Then, the global waveform is reconstructed by a series of neural networks. The experimental results show that the proposed method can obviously alleviate the problems such as the edge delay, overshoot of the distortion waveform, thus it can improve the power estimation accuracy by 32.5%, as well as improve the frequency response characteristics of the high-speed digital acquisition system.
- neural network /
- waveform reconstruction /
- nanosecond pulse /
- waveform distortion /
- frequency response characteristics

HTML全文

图 1 纳秒脉冲测量系统示意图

Figure 1. Schematic diagram of nanosecond pulse measurement system

下载: 全尺寸图片幻灯片

图 2 纳秒脉冲采集系统典型信号

Figure 2. Typical signal of nanosecond pulse acquisition system

下载: 全尺寸图片幻灯片

图 3 波形重建神经网络的训练方法

Figure 3. Training method of waveform reconstruction neural network

下载: 全尺寸图片幻灯片

图 4 训练集波形

Figure 4. Waveforms of the training dataset

下载: 全尺寸图片幻灯片

图 5 预测集不同波形比较

Figure 5. Comparison of different waveforms of the prediction dataset

下载: 全尺寸图片幻灯片

图 6 不同波形的功率估计结果对比

Figure 6. Comparison of power estimation for different waveforms

下载: 全尺寸图片幻灯片

图 7 非典型波形的重建结果

Figure 7. Reconstruction of atypical waveforms

下载: 全尺寸图片幻灯片

表 1 不同波形的功率估计误差

Table 1. Power estimation error of different waveform

waveform type	average of power estimation error/MW	average standard deviation of power estimation error/MW
high-speed digital acquisition waveform	28.26	16.77
reconstructed waveform	0.15	11.32

下载: 导出CSV

表 2 不同信噪比条件下的功率估计误差均值

Table 2. Mean error of power estimation under different signal-to-noise ratio

additional noise standard deviation/V	residual mean of power estimation/MW
additional noise standard deviation/V	high-speed digital acquisition waveform	reconstructed waveform
0	28.26	0.20
0.01	44.83	1.14
0.02	86.31	2.67
0.03	142.07	4.34

下载: 导出CSV

表 3 不同信噪比条件下的功率估计误差标准差

Table 3. Standard deviation of power estimation error under different signal-to-noise ratio

additional noise standard deviation/V	standard deviation of power estimation error/MW
additional noise standard deviation/V	high-speed digital acquisition waveform	reconstructed waveform
0	16.77	11.77
0.01	23.20	14.19
0.02	42.75	19.48
0.03	65.58	25.34

下载: 导出CSV

参考文献(13)

[1]	郭明安, 刘璐, 郭耀军, 等. 一种带输出监测的简单快速高压脉冲源[J]. 现代应用物理, 2022, 13:020204 doi: 10.12061/j.issn.2095-6223.2022.020204 Guo Ming’an, Liu Lu, Guo Yaojun, et al. A simple and fast high voltage pulse source with output monitoring[J]. Modern Applied Physics, 2022, 13: 020204 doi: 10.12061/j.issn.2095-6223.2022.020204
[2]	崔光曦, 李俊娜, 陈旭良, 等. 一种基于Marx发生器的纳秒脉冲实验平台[J]. 现代应用物理, 2022, 13:040402 doi: 10.12061/j.issn.2095-6223.2022.040402 Cui Guangxi, Li Junna, Chen Xuliang, et al. A nanosecond pulse experimental platform based on Marx generator[J]. Modern Applied Physics, 2022, 13: 040402 doi: 10.12061/j.issn.2095-6223.2022.040402
[3]	夏文锋, 张冬晓, 刘启晨, 等. 一种新型高功率轻小型化脉冲驱动源研制[J]. 现代应用物理, 2023, 14:030506 Xia Wenfeng, Zhang Dongxiao, Liu Qichen, et al. A novel lightweight and miniaturized high power pulse drive source[J]. Modern Applied Physics, 2023, 14: 030506
[4]	曾正中. 实用脉冲功率技术引论[M]. 西安: 陕西科学技术出版社, 2003 Zeng Zhengzhong. Introduction to practical pulse power technology[M]. Xi’an: Shaanxi Science and Technology Press, 2003
[5]	陈炜峰, 胡绍朋, 薛冬. 一种基于双传输线的纳秒脉冲源的研制[J]. 科学技术与工程, 2013, 13(27):7992-7996 doi: 10.3969/j.issn.1671-1815.2013.27.012 Chen Weifeng, Hu Shaopeng, Xue Dong. The design of the nanoseconds pulse source based on a pair of transmission lines[J]. Science Technology and Engineering, 2013, 13(27): 7992-7996 doi: 10.3969/j.issn.1671-1815.2013.27.012
[6]	冯远程, 朱旭东, 元勇, 等. 瞬态过电压波形的滤波、重建与拟合[J]. 高电压技术, 2007, 33(7):44-48,71 doi: 10.3969/j.issn.1003-6520.2007.07.010 Feng Yuancheng, Zhu Xudong, Yuan Yong, et al. Filtering, rebuilding and curve fitting of overvoltage transient signals[J]. High Voltage Engineering, 2007, 33(7): 44-48,71 doi: 10.3969/j.issn.1003-6520.2007.07.010
[7]	韩英杰, 孙广生, 严萍, 等. 纳秒脉冲电压的波形重建[J]. 强激光与粒子束, 2004, 16(4):514-516 Han Yingjie, Sun Guangsheng, Yan Ping, et al. Waveform reconstruction of nanosecond pulse voltage[J]. High Power Laser and Particle Beams, 2004, 16(4): 514-516
[8]	曹景阳, 谢树果, 苏东林. 基于最小相位法重建电磁脉冲时域波形[J]. 电波科学学报, 2011, 26(6):1102-1106 Cao Jingyang, Xie Shuguo, Su Donglin. Application of minimum phase method in a pulse measurement[J]. Chinese Journal of Radio Science, 2011, 26(6): 1102-1106
[9]	付佳斌, 卿燕玲, 卫兵, 等. ns脉冲测量中的波形重建[J]. 强激光与粒子束, 2010, 22(11):2759-2762 doi: 10.3788/HPLPB20102211.2759 Fu Jiabin, Qing Yanling, Wei Bing, et al. Waveform reconstruction of nanosecond pulse measurement[J]. High Power Laser and Particle Beams, 2010, 22(11): 2759-2762 doi: 10.3788/HPLPB20102211.2759
[10]	付佳斌, 卿燕玲, 卫兵, 等. 同轴电缆测量纳秒脉冲信号衰减的数字补偿[J]. 强激光与粒子束, 2011, 23(10):2826-2830 doi: 10.3788/HPLPB20112310.2826 Fu Jiabin, Qing Yanling, Wei Bing, et al. Numerical compensation for coaxial cable signal degradation[J]. High Power Laser and Particle Beams, 2011, 23(10): 2826-2830 doi: 10.3788/HPLPB20112310.2826
[11]	陈翔, 魏明, 王向东, 等. 基于RBF神经网络的静电电位动态测试仪波形重建[J]. 计算机测量与控制, 2010, 18(5):1202-1205 Chen Xiang, Wei Ming, Wang Xiangdong, et al. Dynamic tester of electrostatic potential waveform reconstruction based on RBF neural network[J]. Computer Measurement & Control, 2010, 18(5): 1202-1205
[12]	侯媛彬, 杜京义, 汪梅. 神经网络[M]. 西安: 西安电子科技大学出版社, 2007 Hou Yuanbin, Du Jingyi, Wang Mei. Neural network[M]. Xi’an: Xidian University Press, 2007
[13]	史忠植. 神经网络[M]. 北京: 高等教育出版社, 2009 Shi Zhongzhi. Neural networks[M]. Beijing: Higher Education Press, 2009