基于KAN的舰载高功率微波武器目标威胁评估

doi:10.12382/bgxb.2024.0859

摘要/Abstract

摘要：

随着无人机等现代化电子设备在战场上的广泛应用,高功率微波(High Power Microwave,HPM)武器在未来海上防空作战中展现出巨大的潜力。HPM武器与常规舰载防空武器协同防空时的威胁评估成为亟待解决的核心问题,建立基于KAN(Kolmogorov-Arnold Network)的威胁评估模型,通过量化6项核心因素对来袭目标的威胁程度进行评估。仿真实验表明:KAN模型的威胁评估误差保持在0.036以下,平均误差为0.0142,优于传统的BP(Back Propagation)网络模型和FABP(Firefly Algorithm Optimized Back Propagation)网络模型。加入HPM武器后,对相应量化指标进行调整,运用KAN模型预测加入HPM武器后的目标威胁值,目标威胁值平均下降了0.0714。HPM武器的使用显著增强了舰艇的防空能力,为未来防空系统的设计提供了重要指导。

关键词: 高功率微波武器, 威胁评估, Kolmogorov-Arnold网络, 残差激活函数, 稀疏化训练

Abstract:

With the widespread application of modern electronic equipment,such as drones,on the battlefield,the high-power microwave (HPM)weapons have shown great potential in future maritime air defense operations.A key issue that needs to be addressed is the threat assessment when HPM weapons are used in conjunction with conventional shipborne air defense weapons.To tackle this,a threat assessment model based on the Kolmogorov-Arnold network (KAN)is established for evaluating the threat level of incoming targets by quantifying six core factors.Simulation experiments show that the KAN model maintains a threat assessment error below 0.036 with an average error of 0.0142,outperforming traditional BP network models and FABP network models.After incorporating HPM weapons,the corresponding quantitative indicators are adjusted,and the KAN model is used to predict the target threat values.The average target threat value is decreased by 0.0714.The HPM weapons can be usedsignificantly to enhance the air defense capabilities of naval vessels.

Key words: high-power microwave weapon, target threat assessment, Kolmogorov-Arnold network, residual activation function, sparse training

徐洪浩, 曹巍, 章业超, 陈志华. 基于KAN的舰载高功率微波武器目标威胁评估[J]. 兵工学报, 2024, 45(S2): 47-54.

XU Honghao, CAO Wei, ZHANG Yechao, CHEN Zhihua. KAN-based Target Threat Assessment of Shipboard High-power Microwave Weapon[J]. Acta Armamentarii, 2024, 45(S2): 47-54.

图/表 13

图1 目标威胁评估流程

Fig.1 Target threat assessment process

表1 部分威胁评估指标量化

Table 1 Quantifies part of the threat assessment indicators

目标类型	量化	目标速度/ (m·s^-1)	量化	目标干扰能力	量化
中	5	450	2	中	6
大	8	950	5	强	8
小	3	80	1	弱	4

图2 KAN网络结构

Fig.2 KAN network structure

图3 训练集结果

Fig.3 Training set results

图4 训练集误差

Fig.4 Training set error

图5 决定系数

Fig.5 Determination coefficient

图6 均方差

Fig.6 Mean square error

图7 测试集结果

Fig.7 Test set results

图8 测试集误差

Fig.8 Test set error

表2 模型预测值与真实值的平均绝对误差

Table 2 Average absolute errors among predicted values and real values of 3 models

	BP网络	FABP网络	KAN网络
训练集	0.036327	0.016242	0.0003
测试集	0.038803	0.032847	0.0142

表3 高功率微波对电子元器件的毁伤阈值

Table 3 Damage threshold of electronic components caused by high-power microwave

元器件类型	毁伤能量阀值/ (J·cm^-1)	毁伤功率阀值/ (W·cm^-1)
整流管和齐纳二极管	0.5×10^-3~1×10^-3	400~4000
大功率晶体管	1×10^-3~5×10^-3	200~2000
开关二极管	7×10^-5~1×10^-4	30~300
集成电路	1×10^-5~1×10^-4	1~300
微波二极管	7×10^-7~1.2×10^-5	4~100
检波二极管	7×10^-5~1×10^-4	100~1000
点接触二极管	7×10^-6~1×10^-5	7~300

表4 加入HPM武器后部分量化数据

Table 4 Partial quantitative data after adding HPM weapon

目标类型	量化	目标速度/ (m·s^-1)	量化	目标干扰能力	量化
中	4	450	2	中	5
大	8	950	5	强	7
小	2	80	1	弱	3

图9 加入HPM武器目标威胁值

Fig.9 Target threat value after using HPM weapon

参考文献 20

[1]	MESYATS G A, SHPAK V G, YALANDIN M I, et al. Compact high-current accelerators based on the RADAN SEF-303 pulsed power source[C]// Proceedings of the 9th IEEE International Pulsed Power Conference.Albuquerque,NM,US:IEEE, 1993:835-838.
[2]	赵鸿燕, 周丽. 国外高功率微波武器发展研究[J]. 航空兵器, 2023, 30(4):42-48.
	ZHAO H Y, ZHOU L. Research on development of high-power microwave weapons abroad[J]. Aviation Weaponry, 2023, 30(4):42-48. (in Chinese)
[3]	禹化龙, 伍尚慧. 美军定向能武器反无人机技术进展[J]. 国防科技, 2019, 40(6):42-47.
	YU H L, WU S H. Progress of anti-UAV technology with directed energy weapon[J]. National Defense Technology, 2019, 40(6):42-47. (in Chinese)
[4]	陈栋, 许黎明. 高功率微波武器对C⁴ISR系统毁伤效应研究[J]. 装备环境工程, 2010, 7(2):48-50.
	CHEN D, XU L M. Research on damage effect of high power microwave weapon on C⁴ISR system[J]. Equipment Environmental Engineering, 2010, 7(2):48-50. (in Chinese)
[5]	邱千钧, 李梁, 颜燕华, 等. 舰载定向能武器信息系统发展需求研究[J]. 舰船电子工程, 2020, 40(3):15-17.
	QIU Q J, LI L, YAN Y H, et al. Development needs overview of shipboard directional energy weapon in information system[J]. Ship Electronic Engineering, 2020, 40(3):15-17. (in Chinese)
[6]	KOROVIN S D, ROSTOV V V, POLEVIN S D, et al. Pulsed power-driven high-power microwave sources[J]. Proceedings of the IEEE, 2004, 92(7):1082-1095.
[7]	LIU Q Y, WANG J, JUN Y H. Damage evaluation of microwave anti swarm attack based on scoring method[C]// Proceedings of the 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications.Dalian,China:IEEE, 2020:345-350.
[11]	AZIMIRAD E, HADDADNIA J. Target threat assessment using fuzzy sets theory[J]. International Journal of Advances in Intelligent Informatics, 2015, 1(2):57-74.
[12]	YADEGARI M, SEYEDIN S A. Target threat assessment using rule-based joint fuzzy and probabilistic networks[J]. Journal of Electronic Research and Application, 2019, 6(4):61-78.
[13]	LI D, LIU M, ZHANG S. Underwater target threat assessment method based on bayesian network[C]// Proceedings of the 2021 40th Chinese Control Conference.Shanghai,China:IEEE, 2021:3363-3367.
[14]	YU P, LIU G, WEI W, et al. Research on airspace target threat assessment model in sea battlefield based on GA-RBF[C]// Proceedings of the 2019 IEEE 2nd International Conference on Automation,Electronics and Electrical Engineering.Shenyang,China:IEEE, 2019:428-432.
[15]	张峰, 胥文, 吴东岩, 等. 作战仿真中基于BP-Adaboost目标威胁估计研究[J]. 系统仿真技术, 2019, 15(3):180-183.
	ZHANG F, XU W, WU D Y, et al. The research on target threat assessment based on BP-Adaboostin operational simulation[J]. System Simulation Technology, 2019, 15(3):180-183. (in Chinese)
[16]	陈美杉, 钱坤, 李玲杰, 等. 基于ICRITIC-GCN的空战目标威胁评估[J]. 现代防御技术, 2023, 51(2):1-13. doi: 10.3969/j.issn.1009-086x.2023.02.001
	CHEN M S, QIAN K, LI L J, et al. Target threat assessment of air combat based on ICRITIC-GCN[J]. Modern Defence Technology, 2019, 51(2):1-13. (in Chinese)
[17]	LIU Z M, WANG Y X, VAIDYA S, et al. KAN:Kolmogorov-Arnold networks[J]. arxiv preprint arxiv:2404.19756,2024.
[18]	KOLMOGOROV A N. On the representation of continuous functions of several variables by superpositions of continuous functions of a smaller number of variables[M]. Providence,Rhode Island,US: American Mathematical Society.
[19]	BRAUN J, GRIEBEL M. On a constructive proof of Kolmogorov’s superposition theorem[J]. Constructive approximation, 2009, 30:653-675.
[8]	陈凯柏, 高敏, 周晓东, 等. 调频连续波引信高功率微波前门耦合效应研究[J]. 兵工学报, 2020, 41(5):881-889. doi: 10.3969/j.issn.1000-1093.2020.05.007
	CHEN K B, GAO M, ZHOU X D, et al. Research on coupling effect of high-power microwave front-gate in FMCW fuze[J]. Acta Armamentarii, 2020, 41(5):881-889. (in Chinese) doi: 10.3969/j.issn.1000-1093.2020.05.007
[9]	李烨, 郑纯, 马长胜, 等. 基于拦截效率最大化的高功率微波武器系统与中近程防空武器协同作战目标分配模型[J]. 兵工学报, 2023, 44(11):3489-3497. doi: 10.12382/bgxb.2022.0845
	LI Y, ZHENG C, MA C S, et al. Target assignment model for high-power microwave weapon system and medium and short-range air defense weapons in cooperative combat based on maximizing interception efficiency[J]. Acta Armamentarii, 2019, 44(11):3489-3497. (in Chinese)
[10]	DEWANGGA W, SASONGKO R A. Missile threat assessment system using Naive Bayes classifier approach[C]// Proceedings of the 6th International Conference of Science,Technology,and Interdisciplinary Research.Bandar Lampung,Indonesia:.Institute of Physics, 2020.
[20]	王改革. 基于智能算法的目标威胁估计[D]. 北京: 中国科学院大学, 2013.
	WANG G G. Target threat estimation based on intelligent algorithm[D]. Beijing: University of Chinese Academy of Sciences, 2013. (in Chinese)