-
目前,空空、空面、反舰、反坦克等导弹,航空制导炸弹等精确制导武器已成为现在战争中不可或缺的主战装备,已在历次局部战争中被广泛应用并发挥了重要作用。2011年利比亚战争中,以英美法为首的多国联军利用“战斧”巡航导弹、JDAM制导炸弹等多种精确制导武器,对叙利亚政府军的防空导弹、预警雷达及地面部队等目标进行了精确打击,成功达成战术战役目标[1]。2022年俄乌冲突中,俄罗斯动用了包括“伊斯坎德尔”陆基导弹系统、“口径”海基巡航导弹、“匕首”空射高超声速导弹等精确制导武器,对乌克兰的军事指挥所、防空系统、军事基础设施、后勤补给基地等重要军事目标进行中远程精确打击,为有效掌控制空/海权、支援地面作战发挥了重要作用[2]。
表1 列出了俄乌冲突中采用的主流精确制导武器,分析可知:(1)红外、可见光成像等末制导成像体制成为此次俄乌冲突中使用精确制导武器的主流,其优势已经在实战中获得检验,取得了巨大成功;(2)随着战场环境的日渐复杂以及目标对抗性能的不断提高,未来精确制导武器必将走向多模复合寻的末制导体制,而以光电成像为基础的多模制导体制已经成为主流的末制导模式[3]。
表 1 俄乌冲突中参战精确制导武器技战技性能[2]
Table 1. Technical and tactical performance of precision guided weapons in the Russian-Ukrainian conflict [2]
Model Guidance mode Warhead Range/km Speed/Ma Hit accuracy/m 3M-14 "Kalibr" cruise missile Inertial navigation+Satellite navigation+Terrain matching+Active terminal radar guidance 400 kg high bursting disc warhead Submarine-launched: 2 000;
Ship-launched: 1500Max 0.8 5 Iskander-M (9M723) Inertial navigation+Satellite navigation+Optical terminal guidance 400 kg
Cluster/penetrating/anti-personnel explosive warhead480 Max 6 5-10 Iskander-K
(9M728)Inertial navigation+Satellite navigation+Terrain matching+Active terminal radar guidance 400-500 kg high bursting disc warhead 490-500 0.68-0.76 3 Kh-101 air-launched lruise missile Inertial navigation+Satellite navigation+Terrain matching+
IR imaging terminal-guidance400 kg high explosive penetration warhead 4500 0.57-0.8 10 Kh-31P air launched anti-radiation missile Inertial navigation+
Passive terminal radar guidance87 kg high rupture disc kill warhead 110 (High-altitude)
15 (Low-altitude)2-2.9 - Dagger hypersonic air launched missile Inertial navigation+
Terminal radar guidanceNuclear warhead or conventional high explosive warhead 2 000 10 - Kh-29 air-to-ground guided missile TE: Television terminal guidance;
L: Semiactive laser guidance320 kg explosive penetration warhead TE: 3-30;
L: 3-100.8 - "Redoubt" coast ship security complex Inertial navigation+
Active/Passive radar guidance200 kg semi-armor piercing warhead 300 2-2.2 - "Bal" coast ship security complex Inertial navigation+
Active radar guidance145 kg semi-armor-piercing high bursting disc warhead 260 0.8 - KAB-500L laser guided bomb Semi-active laser L: 200 kg high explosive warhead
KL: 200 kg cluster warhead;
LG: 195 kg high explosive warhead9 - L: 7
KL: -
LG: -KAB-500 Kr
television guided bombTelevision guidance Kr: 380 kg penetrate warhead;
OD: 250 kg fuel air warhead;
Kr-E: 80 kg penetrate warhead9 - Kr: 7
OD: 7
Kr-E: -"305" Air-to-ground missile Inertial navigation+Television guidance
Inertial navigation+ Infrared image guidance- - - - KUB-LBA loitering munition Television/Infrared image guidance 3 kg - 0.3 - Javelin anti-tank missile Infrared image guidance Tandem armor-breaking warhead 2 - - Switchblade
–300/600 loitering munitionTelevision/IR imaging dual mode guidance - 9 0.3 - 注:部分数据来自简氏 与此同时,随着人工智能技术的发展与进步,近年来,各军事强国正积极将人工智能技术运用到精确制导武器中,并取得了一定的技术突破,如研制了LRASM反舰导弹[4]、“海上破坏者”、“SPICE-250”精确制导炸弹[5]等智能化武器装备,以提升复杂战场环境的作战效能。在实现精确制导武器智能化的过程中,复杂战场环境自主感知、自动目标截获(ATA)、自动目标识别(ATR)以及自适应导引等成像末制导技术性能的大幅提升依赖于人工智能技术的深度融合应用。因此,进行成像末制导智能化技术与未来发展方向研究,对于紧跟制导模式发展趋势、实现武器作战性能的革命性提升具有重要参考意义。
-
国外典型对海作战武器主要包括空射、岸射、舰射等不同发射形式的反舰导弹,如“飞鱼”、“捕鲸叉”、“花岗岩”等,大多采用雷达末制导,部分采用成像末制导作为辅助末制导方式。随着舰船隐身技术以及有源干扰、无源干扰等“软”杀伤技术的发展,使得对海精确制导武器面临着日益复杂的电磁环境,驱使新一代武器末制导方式由单一探测体制转向包含光电成像探测的多模复合末制导方式。
近年来,为了对海上目标进行精确打击,各国列装的、新研发的采用成像末制导的对海打击精确制导武器如表2所示。
表 2 对海精确制导武器相关性能
Table 2. Performance related to sea precision guided weapons
Model Nation Guidance mode Seeker structure Imaging detector type Detection capabilities Tracking capabilities Level
of intelligenceAGM-84 harpoon anti-ship missile USA Active radar/IR imaging compound guidance - 16-element mercury cadmium telluride detector array Weak Weak Weak LRASM anti-ship missile USA INS/GPS+Data chain+Multi-pattern composite guidance - 256×256 element infrared gaze array General Strong Strong NSM anti-ship missile NO Inertial navigation+GPS+ Terrain matching+Dual band IR imaging terminal-guidance - Medium and long wave two-color infrared gaze array Strong General General Sea Breaker (Missile) ISR Inertial navigation+Terrain matching+IR imaging terminal-guidance - Mid-wave infrared gaze array General Strong Strong 反舰导弹等对海作战武器在成像末制导阶段面临复杂的海背景环境与强电磁对抗环境,其作战目标类型主要为舰船等慢速运动目标。一般地,其成像末制导采用人在回路与智能信息处理融合的方式实现捕获、跟踪,智能信息处理可根据攻击任务和目标特性大致划分为三个阶段:(1) 远距目标截获阶段,主要面临海杂波、海亮带、鱼鳞光、海天线、海地线等复杂背景带来的目标检测问题;(2) 中近距目标跟踪阶段,主要面临海亮带、鱼鳞光、舰船倒影及尾浪、民用船舶、岸岛背景、烟幕对抗等带来的疑似目标与多目标识别、目标遮挡、跟踪点漂移等抗遮挡、抗干扰跟踪问题;(3) 近距目标关键部位识别打击阶段,主要面临目标相对尺度变化、烟幕对抗、目标充满探测器视场等带来的局部关键部位识别、跟踪点选择等精确稳定跟踪问题。相应地,据此分析,传统典型成像末制导信息处理技术原理如图1所示。并在图2、图3中分别展示了三种先进对海作战武器以及舰船目标在不同阶段的红外成像。
图 1 对海目标成像末制导智能信息处理原理框图
Figure 1. Block diagram of intelligent information processing for terminal guidance of imaging to sea targets
远距目标截获阶段,一般采用Top-Hat形态学滤波等单帧背景抑制方法、自适应阈值分割方法提取疑似目标,结合背景噪声、目标灰度分布、面积、梯度及其变化特性抑制背景噪声,同时融合能量累积、管道滤波、动态规划等多帧滤波方法进一步降低虚警。
中近距目标跟踪阶段,现役成像制导反舰导弹等,其成像末制导多采用模式匹配方法[6]。早期采用基于特征相关匹配的自适应跟踪方法,使用目标的几何尺寸、外形轮廓、灰度、位置等简单特征与内存的目标特征模板进行相关处理,实现对目标的跟踪,并依靠相关系数进行跟踪态与搜索态的转换。基于特征相关匹配方法计算速度快、消耗资源少,但是抗背景噪声与抗尺度变化能力鲁棒性比较差。后来,随着弹载计算机等硬件信息处理能力的提高,成像末制导开始使用基于图像模板匹配方法,需要预先存储目标模板,计算效率高,但抗光照、抗背景噪声能力差。随后,发展了基于相关滤波的匹配方法,包括KCF[7]、fDSST[8]、ECO[9]等,抗尺度变化、抗光照变化能力增强,跟踪精度与稳定性均有提升,且计算效率较高,但抗遮挡能力仍然不足。当前,随着深度学习的发展,出现了基于深度学习与相关匹配融合的智能跟踪器,包括SiamFC[10]、SiamRPN[11]、ATOM[12]等方法,抗尺度变化、抗光照变化、抗背景噪声能力大幅增强,也具备了一定的抗遮挡能力,跟踪精度与稳定性均有大幅提升,但计算效率较低且硬件计算能力要求高,同时环境适应性有待提高。
近距目标关键部位识别打击阶段,主要使用基于模型的目标识别来实现对于舰艇关键部位的精确打击[6]。基于模型的目标识别是在统计模式识别的基础上引入模型假设,如垂发装置、指挥舰岛等局部相对于舰船整体的位置关系模型,将分割提取到的目标区域与位置关系模型进行相似性度量,从而最终实现对舰船关键部位的识别与跟踪。
同时,伴随着人工智能的发展,新一代的对海精确制导武器,如LRASM反舰导弹、NSM反舰导弹等更多的采用智能化的多模导引头,通过采用自动目标识别(ATR)、自主导航规划等技术,提升了反舰导弹的突防能力与作战效能,使其具备了智能化的雏形。值得一提的是,2011年以色列的“拉斐尔先进防御系统”有限公司推出的第五代远程、自主、精确制导武器系统“海上破坏者”反舰导弹,支持对静止和运动目标的自动捕获(ATA)和识别(ATR),使用经过训练的人工智能(AI)和机器学习(ML)处理搜索者获得的大数据资源,能够对各种高价值的海上和陆地目标,提供显著的攻击性能。
-
国外典型对地攻击武器主要包括空射、舰射、陆基等不同发射形式的空地导弹、巡航导弹、反坦克导弹等,如“小牛”空地导弹、“战斧”巡航导弹以及“标枪”反坦克导弹等,多采用电视、红外等成像制导作为末制导方式。同时,也发展了多型电视/红外、红外/雷达等多模复合末制导技术的对地武器。为了实现对地面目标的精确打击,各军事强国发展了各种各样的对地精确打击武器,主流的成像末制导对地武器如表3所示。
表 3 对地精确制导武器相关性能
Table 3. Performance related to precision guided weapons to the ground
Model Nation Guidance
modeSeeker
structureImaging
detector
typeDetection
capabilitiesTracking
capabilitiesLevel of
intelligenceMaverick infrared guided air-to-ground missile USA GPS/IR imaging composite guidance type - 16-element mercury cadmium telluride detector array - - Weak SLAM-ER air-to-ground guided missile USA INS/GPS+IR imaging terminal-guidance - 256×256 element infrared gaze focal plane - - General KEPD-350 air-to-ground guided missile GER
SWEINS/GPS+Terrain matching+IBN+IR imaging terminal-guidance Semi-strapdown 256×256 element indium antimonide infrared focal plane FOV: 12° - Weak MMP anti-tank missile Fr TV/uncooled IR dual-mode imaging terminal guidance - - - - General Spike-LR ISR CCD/IR dual mode guidance+Fiber optic or RF data link - - - - General Tomahawk air-launched cruise missile USA IR terminal-guidance - - - - - JDAM IR guided bomb USA IR imaging guidance Strapdown - - - Weak SDB-II guided bomb USA Millimeter-wave/laser/IR imaging three-mode composite guidance - - - - Strong JASSM
air-to-ground guided missileUSA GPS/INS midcourse guidance+IR imaging terminal-guidance - 256×256 element infrared gaze focal plane FOV: 12° - General SPICE-250 guided bomb ISR Television/IR dual-mode imaging terminal guidance - - - - General 空地导弹等对地打击武器在成像末制导阶段面临复杂地面背景环境与强电磁对抗环境,其作战目标类型包括装甲车等快速运动目标、阵地等固定工事。一般地,与反舰导弹类似,其成像末制导采用人在回路与智能信息处理融合的方式实现捕获、跟踪,智能信息处理可根据攻击任务和目标特性大致划分为三个阶段:(1) 远距目标截获阶段,主要面临森林、沙漠、草地、城市、乡村、天地线等复杂地物背景带来的目标检测问题;(2) 中近距目标跟踪阶段,主要面临树林、建筑物、烟尘、民用车辆、伪装、烟幕与诱饵对抗等带来的疑似目标与多目标识别、目标遮挡、跟踪点漂移等抗遮挡、抗干扰跟踪问题;(3) 近距目标关键部位识别打击阶段,主要面临目标相对尺度变化、烟幕对抗、目标充满探测器视场等带来的局部关键部位识别、跟踪点选择等精确稳定跟踪问题。相应地,据此分析,传统典型成像末制导信息处理技术原理如图4所示。并在图5、图6中分别展示了三种先进对地打击武器以及坦克目标在不同阶段的红外成像。
图 4 对地目标成像末制导智能信息处理原理框图
Figure 4. Block diagram of intelligent information processing for terminal guidance of imaging to ground targets
远距目标截获阶段,一般采用背景抑制方法等提取疑似目标,结合背景噪声、目标灰度分布、面积、梯度及其变化特性抑制背景噪声,同时融合能量累积、管道滤波、动态规划等多帧滤波方法进一步降低虚警。
中近距目标跟踪阶段,现役成像制导空地导弹、巡航导弹等,与反舰导弹类似,其成像末制导多采用模式匹配方法[6]。早期的模式匹配方法主要基于图像模板匹配方法[13],利用目标侦察和作战规划等战前保障手段获取目标区的测绘信息,利用坐标变换将目标转换为识别点处的前视图像信息,然后提取目标的边缘结构或纹理特征等作为基准图特征,制作为目标模板数据并装订于弹载计算机,在攻击过程中通过对成像导引头获取的实时图像进行对应特征提取,并与预先装订的基准模板进行匹配识别,从而实现对目标的识别定位。该方法计算效率高,但需要事先准备模板,对作战保障能力要求较高。随着KCF等基于相关滤波的匹配方法的发展,使得地面目标跟踪的跟踪精度与稳定性有了较大提升,且实时性较高,但复杂背景下的跟踪鲁棒性仍有较大的提升空间。当前,随着深度学习的发展,SiamFC等基于深度学习的跟踪算法利用神经网络来实现端对端的相关匹配跟踪,对于复杂地面背景环境、目标尺度剧烈变化等具有较高的鲁棒性,并具备一定的抗遮挡能力,但对硬件计算能力要求较高,且泛化能力有待提升。
近距目标关键部位识别打击阶段,主要使用基于模型的目标识别实现对于关键部位的精确攻击[6]。基于模型的目标识别是在统计模式识别的基础上引入模型假设,将分割提取到的目标区域与位置关系模型进行相似性度量,从而最终实现对关键部位的识别与跟踪。
-
国外典型对空目标作战武器主要包括空空、舰空、地空等不同发射形式的空空导弹、防空导弹等,如“AIM-9X”空空导弹、“标准”系列防空导弹等,中近距格斗弹多采用红外成像制导作为末制导方式,中远程空空、防空导弹多采用雷达制导方式。同时,为有效应对空中目标的威胁,各军事强国积极发展了红外双/多波段、红外/雷达等末制导体制的新型空空、防空导弹等精确制导武器,现役主流对空精确制导武器如表4所示。
表 4 对空精确制导武器相关性能
Table 4. Performance related to precision guided weapons against air
Model Nation Guidance
modeSeeker
structureImaging
detector typeDetection capabilities Tracking capabilities Level of
intelligenceAIM-9X
air-to-air missileUSA IR imaging guidance Semi-strapdown 128×128 element medium wave gaze focal plane Detection range:
10-12 km
FOV: $ \pm {90^ \circ }$Off-axis angle:$ \pm {90^ \circ }$
Angle tracking velocity: 1 600 (°)/sWeak IRIS-T
air-to-air missileGER IR imaging guidance Strapdown 128×4 element medium wave line scan Detection range: 25 km
FOV: $ \pm {90^ \circ }$Off-axis angle:$ \pm {90^ \circ }$
Angle tracking velocity: 1 800 (°)/sWeak ASRAAM
air-to-air missileUK IR imaging guidance Strapdown 128×128 element medium wave gaze focal plane Detection range: 25 km
FOV: $ \pm {90^ \circ }$Off-axis angle: $ \pm {90^ \circ }$ Weak Python-5
air-to-air missileISR IR dual-band composite guidance - 320×240 element dual-band focal plane Detection range: 20 km Off-axis angle: $ \pm {100^ \circ }$ Weak "SM-3"
anti-missile interceptorUSA IR dual-band composite guidance - 512×512 element long-wave dual-band infrared focal plane - - Weak Stunner air-defense missile ISR Photoelectric/IR+
Millimeter wave radar dual-mode composite guidance- CCD/320×240 element dual-band focal plane - - Weak 空空导弹等对空目标作战武器在成像末制导阶段面临复杂的背景环境与强电磁对抗环境,其作战目标类型包括战斗机、轰炸机、导弹等高速、超高速运动目标。一般地,其成像末制导采用智能信息处理方式实现自主捕获、自动识别与跟踪,智能信息处理可根据攻击任务和目标特性大致划分为三个阶段:(1) 远距目标截获阶段,主要面临亮云、海亮带、鱼鳞光、海天线、沙漠、天地线等上视或下视复杂背景带来的目标检测问题;(2) 中近距目标跟踪阶段,主要面临亮云、海亮带、鱼鳞光、红外点源/面源诱饵对抗等带来的多疑似目标识别、干扰严重遮挡、跟踪点漂移等抗遮挡、抗干扰跟踪问题;(3) 近距目标关键部位识别打击阶段,主要面临目标相对尺度急剧变化、诱饵对抗、目标充满探测器视场等带来的局部关键部位识别、跟踪点选择等精确稳定跟踪问题。相应地,据此分析,传统典型成像末制导信息处理技术原理如图7所示。并在图8、图9中分别展示了三种先进对空目标作战武器以及空中目标在不同阶段的红外成像。
图 7 对空目标成像末制导智能信息处理原理框图
Figure 7. Block diagram of intelligent information processing for terminal guidance of imaging to air targets
空中目标与地面以及海面等目标打击任务相比,天空背景与地面、海面等背景相比较为简单,但作战目标具有更高的速度与机动性,对成像末制导系统提出更高的实时性要求。
远距目标截获阶段,一般采用Top-Hat形态学滤波等单帧背景抑制方法、自适应阈值分割方法提取疑似目标,结合背景噪声、目标灰度分布、面积、梯度及其变化特性抑制背景噪声,同时融合能量累积、管道滤波、动态规划等多帧滤波方法进一步降低虚警。
中近距目标跟踪阶段,现役成像制导空空导弹以及防空反导拦截弹等,其成像末制导多采用统计模式识别算法[6]。此类算法在目标场景进行分割的基础上,提取图像中每一目标区域的独立可鉴特征,并采取一定的方式进行融合来获得一组最为有效的特征向量;然后利用目标的先验特征,通过空间区分和时间累积实现对空中目标的实时识别与跟踪。早期利用统计模式识别算法,使用灰度均值、最高灰度、面积、长宽比等简单特征,利用最小距离分类准则等简单的判定准则对空中目标进行识别跟踪,算法鲁棒性较差,智能识别水平比较弱。同时,随着弹载计算机处理能力的提升以及机器学习的发展,对空精确制导武器使用了信息熵、角二阶矩、傅里叶描述子等复杂特征,并使用SVM、贝叶斯等传统机器学习方法作为分类器,实现对干扰条件下的空中目标进行识别,提升了目标识别性能以及智能化水平。当前,随着深度学习的发展,SSD[14]、YOLO[15]等目标检测算法为空中目标的识别跟踪提供了基于数据驱动的端到端的方法,经过初步验证可以获得更高的识别性能,同时具有更好的目标分类能力,使得智能目标识别水平获得了较大提升,但面对日益复杂的空中战场环境,仍然有相当多的问题需要克服。
近距目标关键部位识别打击阶段,早期主要采用角点、质心、形心等几何关键点融合确定跟踪点,跟踪点不稳定且抖动较大。目前,随着基于深度学习的关键点检测算法的深入研究,利用深度学习方法可获得以目标局部部位的识别能力以及更精确的跟踪点,将大大提升精确制导武器的命中精度。
图 8 (a) AIM-9X空空导弹;(b) Python-5空空导弹;(c) “标准-3”反导拦截弹
Figure 8. (a) AIM-9X air-to-air missile; (b) Python-5 air-to-air missile; (c) "SM-3" anti-missile interceptor
综上,国外各军事强国为应对不断出现的高性能目标、日益复杂的对抗环境,为确保精确制导武器的命中精度和作战效能,对于成像末制导的发展方向主要包括以下几个方面:
(1)不断提高单模成像制导的战场环境适应性与抗干扰能力。通过不断提升电视CCD、红外焦平面阵列等成像元件的空间分辨率、灵敏度等性能,为成像末制导系统提供更多的目标信息,使得精确制导武器可以更为有效地应对复杂背景及强干扰环境。
(2)发展多模复合制导方式,获取更多维度的信息。一方面,发展双色/多模复合成像制导,在弥补单模成像制导不足的同时,发挥不同成像元件在不同波段和光谱,对不同背景、目标和干扰的选择性探测和抑制的优势,在能量信息、空间信息的基础上引入光谱信息,提升识别探测能力;另一方面,发展成像/雷达复合制导,充分发挥成像制导与雷达制导的优点,弥合二者的缺陷,提升精确制导武器的全气候全天时适应能力以及复杂战场抗干扰性能。
(3)积极将人工智能技术引入成像末制导,提升巨量战场信息的处理与应对能力。随着人工智能的发展,将目标检测、航迹规划等深度学习算法应用到成像末制导中,充分利用海量的战场信息,发挥自主学习与自主推理能力,打破现有的人工设计准则与已知规律的统计处理模式,使得精确制导武器的作战效能获得大幅提升。
Research and prospect of intelligent technology of optoelectronic imaging terminal guidance
-
摘要: 人工智能技术是实现光电成像精确制导武器智能化、提升复杂作战环境作战性能的重要途径,研究与总结国外成像末制导智能化技术发展现状和趋势对我国精确制导武器智能化技术发展具有重要的指导意义。介绍了光电成像精确制导武器在当前局部冲突中的重要性、多模复合成像末制导及智能化技术发展的必然性。概述了全球主流对海、对地、对空等精确制导武器的成像末制导发展现状,分析总结了光电成像末制导不同阶段智能信息处理面临的关键问题、技术原理和智能化水平。总结了现役或在研的智能化武器装备所具有的智能化特征,并依据现有的人工智能技术分析了其智能技术原理与武器智能化功能实现。从作战目标、作战环境、对抗模式、新型作战样式等方面分析了成像制导导弹所面临的未来复杂战场环境与作战需求,以及对成像末制导智能化技术带来的挑战。依据人工智能与人脑视觉认知能力的对比,提出了未来成像末制导技术智能化的六个能力特征,以及将成像末制导智能化划分为功能级智能、系统级单体智能、体系级群体智能技术三个发展阶段,并分析了相应的技术内涵、关键技术与实现功能。Abstract:
Significance In recent years, various military powers are actively applying AI technology to precision-guided weapons, and have made certain technological breakthroughs, such as the development of LRASM anti-ship missiles, "maritime destroyers", "SPICE-250" precision-guided bombs and other intelligent weapons and equipment to improve the operational effectiveness of complex battlefield environments. In the process of realizing the intelligence of precision-guided weapons, the significant improvement of the performance of imaging terminal guidance technologies such as autonomous perception of complex battlefield environment, automatic target acquisition (ATA), automatic target recognition (ATR), adaptive guidance and so on depends on the deep fusion application of artificial intelligence technology. Therefore, the research on the intelligent technology of imaging terminal guidance and its future development direction has important reference significance for following the development trend of guidance mode and realizing the revolutionary improvement of weapon operational performance. Progress Firstly, the development status of typical photoelectric imaging terminal guidance equipment for sea-to-sea, ground-to-ground and air-to-air, the different complex interference environments and target characteristics faced by the terminal guidance process are analyzed. The typical terminal guidance intelligent information processing principles such as small target detection in the long-range target interception stage, identification, tracking and anti-interference in the medium-close target tracking stage, and identification of key parts at the end of the close range in three scenarios are analyzed. Secondly, the development status and intelligent equipment achievements of intelligent technology of terminal-guided weapons in the United States, Israel, Norway and other foreign countries are summarized, and the intelligent technology principles in automatic target recognition, track planning and other aspects are analyzed, as well as the combat requirements of intelligent weapons in the future complex combat mode, including the significant improvement of target survivability, the increasingly complex and changeable task environment, and the increasingly fierce confrontation environment. Finally, this paper proposes several key technologies for the intelligent requirements of future electro-optical imaging terminal guidance weapons: distributed/heterogeneous autonomous collaborative detection capabilities, multi-dimensional information intelligent fusion processing capabilities, battlefield environment awareness and situation understanding capabilities, detection and guidance integration and autonomous decision-making capabilities, self-learning self-evolution self-reasoning capabilities, collaborative identification and collaborative anti-interference capabilities. At the same time, it is proposed to divide the intelligence of imaging terminal guidance into three stages: functional intelligent technology, system-level single intelligent technology, and system-level group intelligent technology. Conclusion and Prospects This paper analyzes the challenges brought by future high-performance targets, complex confrontation environments, multi-task requirements, and new combat modes to the intelligence of imaging terminal guidance technology. Starting from artificial intelligence technology and future combat requirements, six capability feature requirements and three development stages for realizing intelligent imaging terminal guidance are proposed. Through the development analysis of foreign imaging terminal guidance intelligent technology, it provides reference for the development of intelligent technology of photoelectric imaging terminal guidance weapons in China. -
表 1 俄乌冲突中参战精确制导武器技战技性能[2]
Table 1. Technical and tactical performance of precision guided weapons in the Russian-Ukrainian conflict [2]
Model Guidance mode Warhead Range/km Speed/Ma Hit accuracy/m 3M-14 "Kalibr" cruise missile Inertial navigation+Satellite navigation+Terrain matching+Active terminal radar guidance 400 kg high bursting disc warhead Submarine-launched: 2 000;
Ship-launched: 1500Max 0.8 5 Iskander-M (9M723) Inertial navigation+Satellite navigation+Optical terminal guidance 400 kg
Cluster/penetrating/anti-personnel explosive warhead480 Max 6 5-10 Iskander-K
(9M728)Inertial navigation+Satellite navigation+Terrain matching+Active terminal radar guidance 400-500 kg high bursting disc warhead 490-500 0.68-0.76 3 Kh-101 air-launched lruise missile Inertial navigation+Satellite navigation+Terrain matching+
IR imaging terminal-guidance400 kg high explosive penetration warhead 4500 0.57-0.8 10 Kh-31P air launched anti-radiation missile Inertial navigation+
Passive terminal radar guidance87 kg high rupture disc kill warhead 110 (High-altitude)
15 (Low-altitude)2-2.9 - Dagger hypersonic air launched missile Inertial navigation+
Terminal radar guidanceNuclear warhead or conventional high explosive warhead 2 000 10 - Kh-29 air-to-ground guided missile TE: Television terminal guidance;
L: Semiactive laser guidance320 kg explosive penetration warhead TE: 3-30;
L: 3-100.8 - "Redoubt" coast ship security complex Inertial navigation+
Active/Passive radar guidance200 kg semi-armor piercing warhead 300 2-2.2 - "Bal" coast ship security complex Inertial navigation+
Active radar guidance145 kg semi-armor-piercing high bursting disc warhead 260 0.8 - KAB-500L laser guided bomb Semi-active laser L: 200 kg high explosive warhead
KL: 200 kg cluster warhead;
LG: 195 kg high explosive warhead9 - L: 7
KL: -
LG: -KAB-500 Kr
television guided bombTelevision guidance Kr: 380 kg penetrate warhead;
OD: 250 kg fuel air warhead;
Kr-E: 80 kg penetrate warhead9 - Kr: 7
OD: 7
Kr-E: -"305" Air-to-ground missile Inertial navigation+Television guidance
Inertial navigation+ Infrared image guidance- - - - KUB-LBA loitering munition Television/Infrared image guidance 3 kg - 0.3 - Javelin anti-tank missile Infrared image guidance Tandem armor-breaking warhead 2 - - Switchblade
–300/600 loitering munitionTelevision/IR imaging dual mode guidance - 9 0.3 - 注:部分数据来自简氏 表 2 对海精确制导武器相关性能
Table 2. Performance related to sea precision guided weapons
Model Nation Guidance mode Seeker structure Imaging detector type Detection capabilities Tracking capabilities Level
of intelligenceAGM-84 harpoon anti-ship missile USA Active radar/IR imaging compound guidance - 16-element mercury cadmium telluride detector array Weak Weak Weak LRASM anti-ship missile USA INS/GPS+Data chain+Multi-pattern composite guidance - 256×256 element infrared gaze array General Strong Strong NSM anti-ship missile NO Inertial navigation+GPS+ Terrain matching+Dual band IR imaging terminal-guidance - Medium and long wave two-color infrared gaze array Strong General General Sea Breaker (Missile) ISR Inertial navigation+Terrain matching+IR imaging terminal-guidance - Mid-wave infrared gaze array General Strong Strong 表 3 对地精确制导武器相关性能
Table 3. Performance related to precision guided weapons to the ground
Model Nation Guidance
modeSeeker
structureImaging
detector
typeDetection
capabilitiesTracking
capabilitiesLevel of
intelligenceMaverick infrared guided air-to-ground missile USA GPS/IR imaging composite guidance type - 16-element mercury cadmium telluride detector array - - Weak SLAM-ER air-to-ground guided missile USA INS/GPS+IR imaging terminal-guidance - 256×256 element infrared gaze focal plane - - General KEPD-350 air-to-ground guided missile GER
SWEINS/GPS+Terrain matching+IBN+IR imaging terminal-guidance Semi-strapdown 256×256 element indium antimonide infrared focal plane FOV: 12° - Weak MMP anti-tank missile Fr TV/uncooled IR dual-mode imaging terminal guidance - - - - General Spike-LR ISR CCD/IR dual mode guidance+Fiber optic or RF data link - - - - General Tomahawk air-launched cruise missile USA IR terminal-guidance - - - - - JDAM IR guided bomb USA IR imaging guidance Strapdown - - - Weak SDB-II guided bomb USA Millimeter-wave/laser/IR imaging three-mode composite guidance - - - - Strong JASSM
air-to-ground guided missileUSA GPS/INS midcourse guidance+IR imaging terminal-guidance - 256×256 element infrared gaze focal plane FOV: 12° - General SPICE-250 guided bomb ISR Television/IR dual-mode imaging terminal guidance - - - - General 表 4 对空精确制导武器相关性能
Table 4. Performance related to precision guided weapons against air
Model Nation Guidance
modeSeeker
structureImaging
detector typeDetection capabilities Tracking capabilities Level of
intelligenceAIM-9X
air-to-air missileUSA IR imaging guidance Semi-strapdown 128×128 element medium wave gaze focal plane Detection range:
10-12 km
FOV:$ \pm {90^ \circ }$ Off-axis angle: $ \pm {90^ \circ }$
Angle tracking velocity: 1 600 (°)/sWeak IRIS-T
air-to-air missileGER IR imaging guidance Strapdown 128×4 element medium wave line scan Detection range: 25 km
FOV:$ \pm {90^ \circ }$ Off-axis angle: $ \pm {90^ \circ }$
Angle tracking velocity: 1 800 (°)/sWeak ASRAAM
air-to-air missileUK IR imaging guidance Strapdown 128×128 element medium wave gaze focal plane Detection range: 25 km
FOV:$ \pm {90^ \circ }$ Off-axis angle: $ \pm {90^ \circ }$ Weak Python-5
air-to-air missileISR IR dual-band composite guidance - 320×240 element dual-band focal plane Detection range: 20 km Off-axis angle: $ \pm {100^ \circ }$ Weak "SM-3"
anti-missile interceptorUSA IR dual-band composite guidance - 512×512 element long-wave dual-band infrared focal plane - - Weak Stunner air-defense missile ISR Photoelectric/IR+
Millimeter wave radar dual-mode composite guidance- CCD/320×240 element dual-band focal plane - - Weak -
[1] 朱爱平, 蒋琪, 杨磊, 等. 从利比亚战争看精确制导武器在不对称战争中的应用[J]. 飞航导弹, 2011(4): 1-5. doi: 10.16338/j.issn.1009-1319.2011.04.006. [2] 海鹰资讯. 【热点聚焦】俄乌冲突中俄精确打击武器运用研究 [EB/OL]. [2022-11-10]. https://mp.weixin.qq.com/s/lEX6Ab6r5CidXhWoEYW5vw. [3] Chen Xianzhi, Ren Gang, Luo Zhenbao, et al. Research on application technology of advanced photoelectric seeker [J]. Infrared and Laser Engineering, 2021, 50(9): 20210531. (in Chinese) doi: 10.3788/IRLA20210531 [4] 杨宁, 吴艳梅. 智能化的新型反舰导弹LRASM综述[J]. 军事文摘, 2020(23): 39-43. [5] 兰顺正. 第三条道路产物:SPICE精确制导炸弹[J]. 太空探索, 2022(4): 68-71. [6] Chen Xianzhi, Luo Zhenbao, Li Yiqiang, et al. Application of automatic target recognition in image terminal guidance [J]. Infrared and Laser Engineering, 2022, 51(8): 20220391. (in Chinese) doi: 10.3788/IRLA20220391 [7] Henriques J F, Caseiro R, Martins P, et al. High-speed tracking with kernelized correlation filters [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014, 37(3): 583-596. [8] Danelljan M, Häger G, Khan F S, et al. Discriminative scale space tracking [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 39(8): 1561-1575. [9] Danelljan M, Bhat G, Shahbaz Khan F, et al. Eco: Efficient convolution operators for tracking [C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017: 6638-6646. [10] Bertinetto L, Valmadre J, Henriques J F, et al. Fully-convolutional siamese networks for object tracking [C]//Proceedings of European Conference on Computer Vision, 2016: 850-865. [11] Li B, Yan J, Wu W, et al. High performance visual tracking with siamese region proposal network [C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 8971-8980. [12] Danelljan M, Bhat G, Khan F S, et al. Atom: Accurate tracking by overlap maximization [C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019: 4660-4669. [13] Liu Yunfeng, Yang Xiaogang, Qi Naixin, et al. Forward-looking ATR model based on template matching [J]. Ordnance Industry Automation, 2016, 35(7): 4-6. (in Chinese) [14] Liu W, Anguelov D, Erhan D, et al. Ssd: Single shot multibox detector [C]//Proceedings of European Conference on Computer Vision, 2016: 21-37. [15] Redmon J, Divvala S, Girshick R, et al. You only look once: Unified, real-time object detection [C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016: 779-788. [16] 彭绍荣, 许江湖, 胡生亮, 等. NSM反舰导弹反导对抗方法探析[J]. 飞航导弹, 2021(11): 58-63. doi: 10.16338/j.issn.1009-1319.20210114 [17] 侯学隆, 王宗杰, 谢宇鹏. NSM反舰导弹技术性能与作战能力研究[J]. 飞航导弹, 2020(5): 26-33. doi: 10.16338/j.issn.1009-1319.20190124 [18] Ma Chengqian, Xie Wei, Sun Weijie. Research on reinforcement learning technology: A Review [J]. Command Control & Simulation, 2018, 40(6): 68-72. (in Chinese) [19] Zheng Wanli, Yang Ping, Yan Shaoqiang, et al. Analysis on the research status and development trend of military camouflage technology [J]. Modern Defence Technology, 2022, 50(1): 81-86. (in Chinese) [20] Pan Quan, Wang Zengfu, Liang Yan, et al. Basic methods and progress of information fusion (Ⅱ) [J]. Control Theory & Applications, 2012, 29(10): 1233-1244. (in Chinese)