基于贝叶斯反馈云模型的空中目标威胁评估

在现代战争中,空袭与防空作战仍将是主要作战方式之一。对防空方而言,及早发现空袭目标、准确判断空袭意图,是合理分配火力、最大限度发挥自身抗击能力的基本前提。综合考虑了攻防双方及作战环境等因素的影响,建立空中目标威胁评估指标体系。采用云模型理论较好地处理了空袭的模糊性和随机性,通过贝叶斯反馈对专家主观认识差异进行修正,使专家意见更加统一,使威胁评估结果更为简洁合理,便于指挥员快速准确作出决策。     

融合战术与修正物理特征的空中目标战术类型识别

针对敌方飞行器隐身、伪装技术的日趋成熟,致使我方由侦察设备获取的电磁信号严重变形,特别是物理特征在目标类型识别中的作用日趋减弱的现实问题,提出了一种融合目标战术特征修正物理特征证据的目标战术类型识别方法。该方法从目标战术类型识别的军事需求出发,提出对冲突物理特征证据的阶梯式修正规则,以及对冲突证据折扣因子的优化方法,并将修正后的证据重新组合得到识别结果。仿真实例证明该方法在处理目标冲突物理特征证据时具有高准确率的特点。该方法为复杂环境下的目标战术类型识别问题提供一种新的解决思路。     

Bi‐objective missile rescheduling for a naval task group with dynamic disruptions

This paper considers the rescheduling of surface‐to‐air missiles (SAMs) for a naval task group (TG), where a set of SAMs have already been scheduled to intercept a set of anti‐ship missiles (ASMs). In missile defense, the initial engagement schedule is developed according to the initial state of the defensive and attacking units. However, unforeseen events may arise during the engagement, creating a dynamic environment to be handled, and making the initial schedule infeasible or inefficient. In this study, the initial engagement schedule of a TG is assumed to be disrupted by the occurrence of a destroyed ASM, the breakdown of a SAM system, or an incoming new target ASM. To produce an updated schedule, a new biobjective mathematical model is formulated that maximizes the no‐leaker probability value for the TG and minimizes the total deviation from the initial schedule. With the problem shown to be NP‐hard, some special cases are presented that can be solved in polynomial time. We solve small size problems by the augmented ? ‐ constraint method and propose heuristic procedures to generate a set of nondominated solutions for larger problems. The results are presented for different size problems and the total effectiveness of the model is evaluated.     

Path planning for moving target tracking by fixed-wing UAV

For the automatic tracking of unknown moving targets on the ground, most of the commonly used methods involve circling above the target. With such a tracking mode, there is a moving laser spot on the target, which will bring trouble for cooperative manned helicopters. In this paper, we propose a new way of tracking, where an unmanned aerial vehicle (UAV) circles on one side of the tracked target. A circular path algorithm is developed for monitoring the relative position between the UAV and the target considering the real-time range and the bearing angle. This can determine the center of the new circular path if the predicted range between the UAV and the target does not meet the monitoring requirements. A transition path algorithm is presented for planning the transition path between circular paths that constrain the turning radius of the UAV. The transition path algorithm can generate waypoints that meet the flight ability. In this paper, we analyze the entire method and detail the scope of applications. We formulate an observation angle as an evaluation index. A series of simulations and evaluation index comparisons verify the effectiveness of the proposed algorithms.     

基于神经网络和遗传算法对反装甲目标威胁度评估研究

针对威胁度估计问题和装甲车辆在战场上的实际情况,运用神经网络和遗传算法理论探索反装甲目标威胁度评估问题。基于BP神经网络模型的算法,利用神经网络良好的自适应能力和遗传算法强大的全局搜索能力,通过样本数据训练,提高了威胁度估计的准确性和适应性。经过验证该算法能够有效反应出反各种反装甲目标的威胁度,其稳定性、精确性也比较高。     

基于声自导鱼雷发现概率的本艇机动优化指标

针对纯方位被动定位与跟踪,以提高武器打击效能为目的,提出鱼雷武器发现概率最大的本艇机动优化指标。对指标进行寻优,得到本艇最优轨迹,通过对轨迹形状、性质的分析以及与工程机动轨迹进行定位精度、鱼雷武器打击效能的比较,证明了该指标对处于理论研究阶段的本艇机动问题有积极的参考价值。     

多枚反舰导弹协同攻击在线目标分配

针对反舰导弹群协同打击在线目标分配问题，在多枚导弹区域覆盖攻击的基础上，建立多枚导弹在线目标分配模型，通过导弹与导弹之间的即时数据交互能力实现多弹协同攻击在线目标分配。     

基于逆合成孔径成像激光雷达的自旋小目标成像系统

为了获得毫米级自旋小目标的清晰成像,采用逆合成孔径成像激光雷达技术设计了基于距离向数据与方位向数据相融合的图像重建系统。系统采用大带宽、窄线宽光纤激光器配合调制器实现激光脉冲的线性调频,利用光外差原理对回波信号进行采集处理。结合自旋目标的运动特性,给出了含有自旋分量的回波信号函数方程,并将该分量引入传统的R-D算法中实现了对自旋目标的ISAIL二维图像的重建。实验采用毫米级铝条构成被测小目标,通过步进电机及带倾角的转台完成运动及自旋模拟。实验结果显示,当目标固定时,可通过回波能量数据获得一维距离向图像,与被测目标的4个特征点位置一致。当目标运动时,通过数据压缩并代入自旋参量,最后通过R-D算法可以获得可识别的ISAIL二维图像,验证了系统符合自旋小目标成像的设计要求。     

基于QIPF的弱目标检测前跟踪算法

针对低信噪比条件下弱目标检测跟踪问题,提出一种拟蒙特卡罗智能粒子滤波检测前跟踪算法(Quasi-Monte Carlo Intelligent Particle Filter Track Before Detect,QIPF-TBD)。首先,该算法采用拟蒙特卡罗技术改善探测空间中粒子分布的均匀性;其次,通过对更新阶段的粒子进行交叉变异等操作,提高粒子重采样之后的多样性。与同类算法的仿真分析表明,所提方法能有效改善低信噪比目标的检测概率和跟踪精度。     

基于多通道卷积神经网络的磁性舰船目标运动参数估计

传统磁性目标运动估计效果依赖于目标的初始状态信息,为克服这一缺陷,建立磁性运动目标三分量投影模型,并据此生成磁性舰船运动目标在运动速度、航向、信噪比等参数变化情况下的10类目标的训练数据集、验证数据集以及测试数据集。设计多通道卷积神经网络,对目标的正横距离和运动速度进行估计,并比较和分析了不同的学习方式和激活函数对网络性能的影响。结果表明:Adam+tanh组合方式的估计性能要优于其他组合方式,而且对磁性目标运动参数的估计效果比较精确,此方法相较于卡尔曼滤波、粒子滤波等估计算法的优越性在于运算复杂度低以及参数估计不需要目标初始状态信息。