基于BAS-BPNN的调频无线电引信目标与扫频干扰识别方法

doi:10.12382/bgxb.2022.0248

摘要/Abstract

摘要：

针对调频无线电引信在复杂电磁战场环境对抗调幅扫频式干扰能力不足的问题,提出一种基于频域信息熵、范数熵和倒频谱熵的调频无线电引信目标识别方法。基于目标和调幅扫频干扰作用下的调频无线电引信检波端输出信号,提取频域信息熵、范数熵和倒频谱熵构建特征矩阵,并利用天牛须搜索(BAS)算法对反向传播神经网络(BPNN)初始权重值和阈值进行优化,利用优化后的BPNN对目标和调幅扫频干扰信号进行分类识别。实测数据实验结果表明,特征提取方法构成的特征矩阵在目标与干扰之间具备可分性,BAS算法优化获得最优参数的BPNN时,该分类器的识别准确率可以达到99.96%,显著提升了调频无线电引信对抗调幅扫频干扰的能力。

关键词: 调频无线电引信, 熵特征, 目标识别, 反向传播神经网络, 天牛须算法

Abstract:

In order to solve the problem that the FM radio fuze is unable to counter AM frequency sweep jamming signals on battlefield in a complex electromagnetic environment, a target recognition method based on frequency domain information entropy, norm entropy and cepstrum entropy is proposed. Based on the output signal of the FM radio fuze under the action of the target and AM frequency sweep jamming signal, the frequency information entropy,norm entropy and cepstrum entropy are extracted to construct the feature matrix. The BAS algorithm is used to optimize the initial weight values and threshold of the back propagation neural network (BPNN). Then the optimized BPNN is used to classify and recognize the target and AM frequency sweep jamming signal. The experimental results with the measured data show that the feature matrix formed by feature extraction has separability between the target and the jamming signal. When the BPNN with optimal parameters is obtained by the optimization of the BAS algorithm, the recognition accuracy of the classifier can reach 99.96%, which significantly improves the ability of the FM radio fuze to counter AM frequency sweep jamming signals.

Key words: FM radio fuze, entropy features, target recognition, BPNN, beetle antennae search algorithm

中图分类号:

TJ43⁺

刘冰, 郝新红, 周文, 杨瑾. 基于BAS-BPNN的调频无线电引信目标与扫频干扰识别方法[J]. 兵工学报, 2023, 44(8): 2391-2403.

LIU Bing, HAO Xinhong, ZHOU Wen, YANG Jin. Recognition Method of Target and Sweep Jamming Signal for FM Radio Fuze Based on BAS-BPNN[J]. Acta Armamentarii, 2023, 44(8): 2391-2403.

图/表 17

图1 目标和不同干扰信号作用下引信检波端输出信号

Fig.1 Fuze output signal under the action of the target and different jamming signals

图2 引信检波端信号预处理

Fig.2 Fuze output signal preprocessing

图3 范数熵P值与差异显著性检验输出值的关系

Fig.3 Relationship between norm entropy P value and output of significance test of difference

图4 频域信息熵、范数熵和倒频谱熵概率密度分布

Fig.4 Frequency information entropy, norm entropy and cepstrum entropy probability density distributions

图5 目标和干扰信号频域信息熵、范数熵和倒频谱熵箱型图

Fig.5 Frequency information entropy, norm entropy and cepstrum entropy box plots of target and jamming signal

图6 BAS算法搜索Ackley函数最小值结果

Fig.6 Using BAS algorithm for minimum value result of Ackley function

图7 BAS算法搜索Cross-In-Tray函数最小值结果

Fig.7 Using BAS algorithm for minimum value result of Cross-In-Tray function

图8 BAS确定最优网络初始参数流程图

Fig.8 Flow chart of BAS determining initial parameters of optimal network

图9 实验场景示意图

Fig.9 Schematic diagram of experimental scene

表1 实验参数设置

Table 1 Experimental parameters setting

部件	参数	参数设置
	载波频率	f₀
引信	调制频偏	ΔF
	调制信号频率	f_m
干扰机	扫频带宽	f₀±ΔF
	扫频步长	$2 Δ F N$

表1 实验参数设置

Table 1 Experimental parameters setting

部件	参数	参数设置
	载波频率	f₀
引信	调制频偏	ΔF
	调制信号频率	f_m
干扰机	扫频带宽	f₀±ΔF
	扫频步长	$2 Δ F N$

图10 Sigmoid激活函数

Fig.10 Sigmoid activation function

表2 识别准确率随迭代次数变化

Table 2 Recognition accuracy varying with the number of iterations

迭代次数	识别准确率/%	迭代次数	识别准确率/%
1	58.13±2.2316×10^-16	16	99.71±0.0070
2	58.13±2.2316×10^-16	17	99.72±0.0063
3	58.13±2.2316×10^-16	18	99.73±0.0057
4	58.13±2.2316×10^-16	19	99.84±0.0053
5	58.19±0.0026	20	99.86±0.0050
6	59.38±0.0241	21	99.92±0.0023
7	64.70±0.0590	22	99.91±0.0026
8	76.04±0.0949	23	99.93±0.0026
9	84.18±0.0835	24	99.94±0.0026
10	92.72±0.0621	25	99.95±0.0019
11	94.74±0.0461	26	99.96±0.0016
12	98.09±0.0211	27	99.93±0.0026
13	98.83±0.0142	28	99.94±0.0024
14	99.43±0.0098	29	99.96±0.0015
15	99.50±0.0104	30	99.96±0.0016

图11 不同迭代次数识别准确率和标准偏差

Fig.11 Recognition accuracy and standardn deviation of different iterations

表3 识别准确率随训练样本数占比变化

Table 3 Recognition accuracy varying with the proportion of training samples

训练样本占比	1/10	1/9	1/8	1/7	1/6	1/5	1/4	1/3	1/2
识别准确率/%	99.78± 0.0077	99.81± 0.0056	99.85± 0.0045	99.90± 0.0023	99.84± 0.0037	99.94± 0.0015	99.89± 0.0030	99.92± 0.0022	99.93± 0.0027

图12 不同训练样本占比识别准确率和标准偏差

Fig.12 Recognition accuracy and standardn deviation of different training sample rates

表4 不同算法识别准确率对比

Table 4 Recognition accuracy varying with different methods

训练样本占比	迭代次数	识别准确率/%
训练样本占比	迭代次数	BAS-BPNN算法	BPNN算法
	15	99.50±0.0104	97.8619±0.0097
2/3	20	99.86±0.0050	98.3886±0.0042
	25	99.95±0.0019	98.5979±0.0013
	15	99.29±0.0123	97.6666±0.0116
1/3	20	99.78±0.0050	98.1846±0.0047
	25	99.94±0.0016	98.5865±0.0006

图13 不同算法识别准确率

Fig.13 Recognition accuracy of different methods

参考文献 20

[1]	崔占忠, 宋世和, 徐立新. 近炸引信原理[M]. 第3版. 北京: 北京理工大学出版社, 2009: 76-100.
	CUI Z Z, SONG S H, XU L X. Principle of proximity fuze[M]. 3rd ed. Beijing: Beijing Institute of Technology Press, 2009:76-100. (in Chinese)
[2]	赵惠昌. 无线电引信设计原理与方法[M]. 北京: 国防工业出版社, 2012: 35-62.
	ZHAO H C. Principle and method of radio fuze design[M]. Beijing: National Defense Industry Press, 2012:147-162. (in Chinese)
[3]	汪仪林, 马秋华, 张龙山, 等. 引信智能化发展构想[J]. 探测与控制学报, 2018, 40(3): 1-5.
	WANG Y L, MA Q H, ZHANG L S, et al. The Development visualization of fuze intell-igentize[J]. Journal of Detection & Control, 2018, 40(3): 1-5. (in Chinese)
[4]	汪仪林, 张龙山, 马秋华. 从美军标看我国引信安全性的差距[J]. 探测与控制学报, 2019, 41(3): 1-5.
	WANG Y L, ZHANG L S, MA Q H. The deficiencies of domestic fuze safety refering to U.S. millitary standards[J]. Journal of Detection & Control, 2019, 41(3): 1-5. (in Chinese)
[5]	AZADEGAN B, SOLERMANI H, SOLEIMANI M. Resistance against spot jamming by using pulse compression based on quadru-phase and frequency codes in radar systems[J]. Digital Signal Processing, 2020, 99(4):102679-102691. doi: 10.1016/j.dsp.2020.102679 URL
[6]	代健, 李泽, 郝新红, 等. 基于目标联合特征提取的脉冲多普勒引信抗干扰方法[J]. 兵工学报, 2019, 40(2): 225-233. doi: 10.3969/j.issn.1000-1093.2019.02.001
	DAI J, LI Z, HAO X H, et al. Anti-jamming method for pulse doppler fuze based on joint feature extraction of target signal[J]. Acta Armamentarii, 2019, 40(2): 225-233. (in Chinese) doi: 10.3969/j.issn.1000-1093.2019.02.001
[7]	KIM J, LEE S, KIM Y H, et al. Classification of interference signal for automotive radar system with convolutional neural network[J]. IEEE Access, 2020, 8(9):176717-176727. doi: 10.1109/Access.6287639 URL
[8]	YAN Z S, HAO X H, LI R, et al. Target re-cognition technology of pulse doppler radio fuze based on bandwidth characteristics for anti-chaff jamming[C]// Proceedings of 2019 IEEE Asia-Pacific Conference on Applied Electromagnetics. Malacca, Malaysia: IEEE, 2019.
[9]	RAI P K, IDSOE H, YAKKATI R R, et al. Localization and activity classification of unmanned aerial vehicle using mmWave FMCW radars[J]. IEEE Sensors Journal, 2021, 21(14):16043-16053. doi: 10.1109/JSEN.2021.3075909 URL
[10]	ZHAO S S, ZHANG L R, ZHOU Y, et al. Signal fusion based algorithms to discriminate between radar targets and deception jamming in distributed multiple radar architectures[J]. IEEE Sensors Journal, 2015, 15(11):6697-6706. doi: 10.1109/JSEN.2015.2440769 URL
[11]	郝新红, 杜涵宇, 陈齐乐. 调频引信粗糙面目标与干扰信号识别[J]. 北京航空航天大学学报, 2019, 45(10):1946-1955.
	HAO X H, DU H Y, CHEN Q L. Rough surface target and jamming signal recognition of FM fuze[J]. Journal of Beijing University of Aeronautics and Astronautics, 2019, 45(10): 1946-1955. (in Chinese)
[12]	刘歌, 张国毅, 田榛熔. 基于时频图像三维熵特征的雷达信号识别[J]. 火力与指挥控制, 2016, 41(12):169-173.
	LIU G, ZHANG G Y, TIAN Z R. Radar signal recognition based on three dimensional entropy features of time-frequency image[J]. Fire Control & Command Control, 2016, 41(12): 169-173. (in Chinese)
[13]	谢平, 杜义浩, 黄双峰. 基于信息熵的复杂机械系统非线性特征提取方法研究[J]. 振动与冲击, 2008, 27(7):135-137.
	XIE P, DU Y H, HUANG S F. Research on non-linear feature extracting method of complex mechanical system based on information entropy. Journal of Vibration and Shock, 2008, 27(7):135-137. (in Chinese)
[14]	张葛祥, 胡来招, 金炜东. 基于熵特征的雷达辐射源信号识别[J]. 电波科学学报, 2005, 20(4): 30-35.
	ZHANG G X, HU L Z, JIN W D. Radar emitter signal recognition based on entropy features[J]. Chinese Journal of Radio Science, 2005, 20(4): 30-35. (in Chinese)
[15]	GUAN X M, MA J X, ZHANG W D. Detection and classification on amateur drones based on cepstrum of radio frequency signal[J]. Transactions of Nanjing University of Aeronautics and Astronautics, 2021, 38(4): 597-606.
[16]	JIANG X Y, LI S. BAS: beetle antennae search algorithm for optimization problems[J]. International Journal of Robotics and Control, 2018, 1(1): 2577-2582.
[17]	JIANG X Y, LI S. Beetle antennae search without parameter tuning (BAS-WPT) for multiobjective optimization[J]. Filomat, 2020, 34(15): 5113-5119. doi: 10.2298/FIL2015113J URL
[18]	周志华. 机器学习[M]. 北京: 清华大学出版社, 2016: 97-114.
	ZHOU Z H. Machine learning[M]. Beijing: Tsinghua University Press, 2016:97-114. (in Chinese)
[19]	ROTH M W. Survey of neural network technology for automatic target recognition[J]. IEEE Transcations on Neural Networks, 1990, 1(1): 28-43.
[20]	SCHUESSLER C, HOFFMANN VOSSIEK M. Super-resolution radar imaging with sparse arrays using a deep neural network trained with enhanced virtual data[J]. IEEE Journal of Microwaves, 2023, 3(3):980-993. doi: 10.1109/JMW.2023.3285610 URL