基于自适应深度网络的无人机目标跟踪算法

针对无人机（UAV）视频中目标易受到遮挡、形变、复杂背景干扰等问题，提出一种基于自适应深度网络的无人机目标跟踪算法。首先，基于主成分分析（PCA）和卷积神经网络（CNN）算法，设计3阶的自适应深度网络进行目标特征提取，该网络对图像的H、S、I通道分别进行主成分分析学习，将得到的特征向量输入网络进行分层卷积，优化了网络结构，提高了网络的收敛速度和精度。其次，将目标深度特征输入核相关滤波算法进行目标跟踪，通过分析相邻2帧图像的变化率，采用分段自适应调整学习率的算法进行目标模板更新，有效地改善目标遮挡问题。仿真实验结果表明，该算法有效地避免了复杂因素干扰导致的跟踪精度下降，具有较好的鲁棒性，相较于全卷积跟踪（FCNT）算法平均跟踪精度提高了9.62%，平均跟踪成功率提高了11.9%。     

Vision-based bio-inspired guidance law for small aerial vehicle

During predation, a flying insect can form a stealth flight path. This behavior is called motion camouflage. Based on the study results of this behavior, the perception and neurology of flying insects, a novel bio-inspired guidance law is proposed for the terminal guidance for small aerial vehicle with charge-coupled device imaging seekers. The kinematics relationship between a small aerial vehicle and target is analyzed, and a two-dimensional guidance law model is established by using artificial neural networks. To compare with the proportional guidance law, the numerical simulations are carried out in the vertical plane and in the horizontal plane respectively. The simulation results show that the ballistic of the small aerial vehicle is straighter and the normal acceleration is smaller by using the bio-inspired guidance law than by using the proportional guidance law. That is to say, the bio-inspired guidance law just uses the information of the target from the imaging seeker,but the performance of it can be better than that of the proportional guidance law.     

一种星-箭连接动态界面力的深度学习反演方法

针对于星-箭连接动态界面力无法通过力传感器直接测量,且典型时域动载反演方法难以准确计算界面力的时域变化等难点,提出了基于长短时记忆(LSTM)神经网络的星-箭界面力深度学习反演方法。首先通过卫星地面测试试验得到数据依据,以卫星主体结构的加速度测量数据为输入层,以星-箭界面力测量数据为输出层,利用LSTM神经网络建立输入和输出间的反演映射关系模型,实现卫星在发射过程中较高精度的界面力反演。进而,设计并开展了某典型卫星结构的正弦扫频和随机振动实验,测试LSTM界面力反演方法的可行性。结果分析可知,所提出的基于LSTM深度学习反演方法能够精确地获得动态界面力时程数据,两项性能指标均优于目前典型的载荷反演方法。     

改进PointNetLK的点云智能配准与位姿图优化方法

针对空间在轨服务任务中的非合作目标相对位姿测量问题,提出一种目标可测部位点云的智能配准方法。首先,通过Straight Through滤波算法对半物理仿真平台采集得到的点云进行目标提取,以消除背景数据等杂乱信息;其次,改进PointNetLK神经网络点云配准算法,将提取后的点云数据作为输入,从而获得初步配准结果,解决非合作目标先验信息缺失导致的无法配准问题;最后,建立基于位姿图的优化模型,以降低配准误差,提高配准精度。实验结果表明,与传统迭代最近点(ICP)算法相比,配准综合误差从6.3598降低到1.7291,精度提高约 72.81% 单次耗时从33.16 s降低到4.2 s,效率提升约87.33%,与当前SM ICP等其他算法相比,也具有一定的优势。     

航天器轨道追逃博弈多阶段强化学习训练方法

针对航天器轨道追逃博弈问题,提出一种多阶段学习训练赋能方法,使得追踪星在终端时刻抵近逃逸星的特定区域,而逃逸星需要通过轨道机动规避追踪星。首先,构建两星的训练策略集,基于逻辑规则设计追踪星和逃逸星的机动策略,通过实时预测对方的终端位置,设计己方的期望位置和脉冲策略,显式给出追逃策略的解析表达式,用于训练赋能;其次,为提升航天器的训练赋能效率及应对未知环境的博弈能力,提出一种基于强化学习技术多模式、分阶段的学习训练方法,先使追踪星和逃逸星分别应对上述逻辑规则引导下的逃逸星和追踪星,完成预训练;再次,开展二次训练,两星都采用邻近策略优化(PPO)策略进行追逃博弈,在博弈中不断调整网络权值,提升决策能力;最后,在仿真环境中验证提出的训练方法的有效性,经过二次训练后,追踪星和逃逸星可有效应对不同策略驱动下的对手,提升追逃成功率。     

基于GA-ELM的飞行载荷参数识别

针对用复杂飞行数据识别飞行载荷时的精度低、速度慢等问题,提出一种结合遗传算法(GA)和极限学习机(ELM)的GA-ELM模型。该模型使用ELM神经网络作为计算核心,用遗传算法产生ELM网络输入层到隐含层的权值矩阵和隐含层偏移量;用GA-ELM模型对飞行数据进行识别,并与BP神经网络和原始ELM神经网络的识别结果进行对比。结果表明:GA-ELM模型是一种有效且高精度的飞行载荷参数识别方法。     

Envelope protection for aircraft encountering upset condition based on dynamic envelope enlargement

Upset condition encountered by an aircraft in flight could pose great threat to flight safety, which is of chief importance in civil aviation. The causal factors have the nonlinear and multiple characteristics, and as a result the conventional envelope protection system cannot successfully do with the condition. So dynamic envelope based on differential manifold theory, which can take more coupling factors into account, is proposed as a basis to design a novel envelope protection system. Then the relationship between the dynamic envelope and the control coefficient or pilot command is obtained, and the result shows that the dynamic envelope can be enlarged with the change of control coefficient. Furthermore, quantification of flight security is realized via defining relative distance between flight state and dynamic envelope, which can detect whether there is a risk for an aircraft in flight. Finally, an envelope protection system based on dynamic envelope enlargement is proposed on the basis. NASA’s Generic Transport Model encountering hazard gust wind in climbing phase is taken as an example to verify the system’s feasibility. The result shows that the system can give a better operation encountering upset condition and to a certain extent reduce the number of accidents or incidents.     

应用三维粒子动态分析仪测量三分支联接流场

应用三维粒子动态分析仪 (3D PDA)测得了矩形截面三分支联接的三维速度分布。在测量中 ,经过理论分析与对比试验 ,选择了蚊香烟雾作为散射粒子。实验结果表明 ,在只有支管进气的情况下 ,在总管的封闭端和支管下游总管壁面处产生了回流 ;以及气流从支管进入总管后由于截面扩张在总管横截面上所诱发的二次流 ,并且二次流现象只发生在支管下游附近的总管横截面上。     

MBD环境下计算机辅助零件分类技术

针对MBD环境下的零件分类问题，研究MBD信息提取、零件编码转换和深度学习分类算法技术，并创建MBD-计算机辅助零件分类系统（MBD-CAPC），实现从输入MBD模型到输出零件分类结果的自动化。     

二次奖罚学习自动机

研究了奖罚型学习自动机的一种非线性强化算法。与线性的奖罚模型（ＬＲＰ）不同，新模型的行动选择概率的更新函数为二次的。这使得该模型的学习性能优于ＬＲＰ，且对不同的环境，其具有不同的行为和特点。