The unmanned and intelligent transformation of military equipment has triggered great revolutions in battlefield morphology and combat patterns. Among them, unmanned autonomous vehicles have played an important role in military operations by participating in various tasks such as reconnaissance, target striking, and transportation. Among its structures, the powertrain is a core component of unmanned autonomous vehicles. The operation and maintenance of the powertrain are essential for the equipment's safety and reliability. And the quality of operation and maintenance directly determines the durability of unmanned autonomous vehicles. Therefore, this paper analyzed the realistic requirements of state control of powertrain and summarized technical difficulties as three aspects, including lack of data, lack of calculation, and lack of algorithm evolution. Through the systematic combing of the digital twin concept and the current situation, it is found that the digital twins can play three beneficial roles, including optimizing the layout and selection of physical space sensors, realizing the effective fusion of information space and physical space, and improving generalization and adaptability. An architecture of the digital twin system for unmanned autonomous vehicles powertrain is proposed, in which the key technologies are condensed, such as the overall design technology of the system, multi-dimensional and multi-state data acquisition, and analysis for virtual space mapping, the model construction of the digital twin system for condition monitoring and control, the fault diagnosis and health assessment based on the digital twin, and the decision-making of unmanned autonomous vehicles maintenance based on the health assessment.

针对未来信息化战争条件下航天复杂装备维修保障面临的型号种类多、模块可视化程度低、战时保障前后端衔接不足、人员训练量化评估考核手段缺乏、维修大数据难以挖掘和复用等瓶颈问题,以典型武器装备为对象,提出一种基于增强现实技术、数字孪生内核驱动的维修保障与人员训练系统框架,阐述了数字孪生模型构建、故障在线诊断、维修流程诱导、人员动态评估等关键环节的技术途径,论述了航天复杂装备辅助维修系统的开发与集成要点并展示了原型系统,为实现靠前保障、快速保障提供了可行的技术路径.

为了提高脑控虚拟现实(VR)飞行模拟驾驶的准确率,提出一种基于桌面式虚拟现实技术的立体视觉刺激脑机接口(BCI)系统.该系统运用自主研发的桌面式虚拟现实视觉刺激子系统提供平面视觉刺激和立体视觉刺激2种刺激模式.结合Emotive EPOC+脑电设备采集用户的稳态视觉诱发电位(SSVEP)信号并传输至OpenVIBE脑电处理模块,利用OpenVIBE脑电处理模块对脑电信号进行滤波处理、特征提取和BCI分类器训练,然后运用训练好的BCI分类器在线对脑电信号进行采集和分类并实时转换为飞行控制指令,实现对虚拟现实(VR)飞行模拟器的在线操控.研究结果表明:在刺激频率为8.57,10,12和15 Hz及采集时间窗口为1.5 s的条件下,与平面视觉刺激相比,立体视觉刺激模式下的脑机接口分类器的平均准确率提升了6.5％,由此可见,立体视觉刺激能够诱发用户产生更具激励性的脑电响应,有利于提高脑控飞行模拟驾驶的性能.

Advancement of brain-computer interface (BCI) has shown its applications in various scenarios, including flight control. Flight simulator is a crucial part for aircraft design or experiment. Desktop virtual reality (VR)-based flight is a perfect choice for overcoming existing problems in head-mounted VR flight simulations, such as dizziness and isolation, which make interaction and sharing very difficult. In this paper, a BCI based on the steady-state visual evoked potential paradigm and a VR flight simulator were developed and integrated. The performance of the developed system was evaluated quantitatively for comparative studies. Experimental results show that the developed system is very convenient and suitable for VR flight simulations. The average operating accuracies with plane and VR visual stimuli are 81.6% and 86.8%, respectively. The VR visual stimuli can improve the average operating accuracy by 5.2% compared with the plane visual stimuli.

碳纤维增强复合材料(CFRP)广泛应用在航空航天等领域中,其内部缺陷易引发灾难性的事故,X射线成像是CFRP缺陷检测的常用手段.为了有效减少图像背景对环状CFRP X射线图像缺陷检测性能的影响,提出了一种结合LeNet-5卷积神经网络和图像变换的环状CFRP图像缺陷检测新方法.首先对环状CFRP的X射线图像进行极坐标变换,然后提取变换图像中的感兴趣区域并对其进行分块构成LeNet-5网络训练和测试的数据集,最后根据图像块的二分类结果得到缺陷的局部区域,实现缺陷检测.实验结果表明,所提方法能显著提高缺陷检测性能,与利用原始图像对LeNet-5进行训练相比,该方法使得缺陷检测的召回率、查准率和F1值分别提高了11.02％、38.60％和25.02％.

为了缩短航天维修人员的培训时间,提高维修人员的操作熟练度,设计开发了一款基于桌面虚拟现实设备的航天器维修仿真系统,摆脱了头盔式虚拟现实设备笨重、与现实世界隔绝、易头晕目眩、难以分享等问题;提出了基于XMind思维导图及SqlServer数据库的装配管理方案,实现装配规则编辑的可视化,使得装配与拆解灵活便捷.

Spectral CT can separate basis materials, and thus it can provide information on material characterization and quantification. Such information can benefit various clinical applications. However, the presence of non-ideal effects in X-ray imaging systems limits the accuracy of basis images. To achieve high accuracy of material decomposition and high quality of basis images, a novel direct iterative basis material image reconstruction based on maximum a posteriori expectation–maximization algorithm (MAP-EM-DD) is proposed. Furthermore, by incorporating polar coordinate transformation into MAP-EM-DD, MAP-EM-PT-DD is proposed. The iterative formulas of MAP-EM-DD and MAP-EM-PT-DD are derived. To evaluate the proposed methods, a simulated cylinder phantom with inserts that contain polyethylene, hydroxyapatite, salt water, air, and aluminum is established. The methods are quantitatively evaluated for comparative studies. Results show that the proposed methods can remarkably reduce the noise of basis images and error of material decomposition and improve the contrast-to-noise ratios (CNRs) of each material-specific region. Compared with the image domain material decomposition based on FBP algorithm (FBP-IDD), MAP-EM-DD can reduce the noise levels of basis images ranging from 57.4 to 63.6% and the error levels of each material-specific region from 31.7 to 62.1%. Simultaneously, the CNRs of each material-specific region are improved ranging from 63.8 to 237.3%. Compared with MAP-EM-DD, MAP-EM-PT-DD can reduce the noise levels of basis images ranging from 21.4 to 23.6%, the error levels of each material-specific region ranging from 1.9 to 36.3%, and the reconstruction time of basis images by 14.1%.

为了解决经狭窄腔对内部目标进行多自由度大范围检测的难题,设计了一种新型线驱动连续型机器人.首先运用几何分析对该机器人进行建模,研究了单组关节的驱动空间、关节空间和操作空间之间的运动学映射定量关系,并对其工作空间进行了分析.针对2组关节协同运动时存在的耦合问题,提出了一种新的运动学解耦算法,并对线驱动连续型机器人单组关节和2组关节运动学特性进行了仿真研究.结果表明:所设计的连续型机器人能够经狭窄腔实施大范围空间的多自由度检测作业,具有良好的弯曲性能,其单组关节最大作业半径为99.33 mm;所提出的解耦算法简明有效,为经狭窄腔对内部目标进行多自由度大范围检测的线驱动连续型机器人系统的研制奠定了理论和技术基础.

为了提高基于Kinect相机的呼吸运动监测精度,提出了一种结合目标检测与目标跟踪算法的实时呼吸运动监测方法.该方法首先基于标记板形状特征利用霍夫圆检测算法进行目标检测,然后基于彩色图像结合Camshift目标跟踪算法实现单个或多个目标的跟踪,依据跟踪的结果获取目标位置的深度信息,并对获取的深度信息进行降噪处理.在此基础上,开发了一款基于Kinect相机的呼吸运动精准监测软件系统.实验结果表明,所提出的实时呼吸运动监测方法可有效提高呼吸运动监测的精度,与现有的基于深度图像获取呼吸运动数据方法相比,在体态偏移状态下数据采集准确度由47.9％提升至94.1％.同时,该方法具有良好的可视化效果.

为了提高双能CT基材料分解的精度,降低基材料图像的噪声,提出了基于MAP-EM算法的直接迭代基材料分解方法.结合MAP-EM算法,推导出基材料分解直接迭代求解公式,基于双能投影数据集直接重建基材料分解图像,并对该方法的性能进行了评价和分析.仿真结果表明,所提方法可显著降低分解误差和基材料图像噪声,提高对比噪声比.与基于FBP算法的图像域基材料分解方法相比,该方法可使基材料图像中各材料区域的噪声水平下降57.42％ ～ 63.64％,分解误差水平降低31.72％ ～62.14％,对比噪声比提高1.37％ ～223.17％.

Dual energy computed tomography (DECT) plays a significant role in medical application and public security. The accuracy of the decomposition relies much on the quality of the reconstructed images. Images reconstructed by traditional methods generally suffer from significant noise, leading to the low accuracy of the material decomposition and identification. In order to solve the above problem, the maximum a posteriori expectation maximization (MAP-EM) is employed to reconstruct the CT images for the material decomposition, and the performance of the material decomposition is evaluated. A dual source DECT system with 140/80 kVp are simulated by Geant4. The MAP-EM algorithm is used to reconstruct the images from the collected projection at two different tube voltages. And then the basis material decomposition coefficient images are obtained with the basis material decomposition coefficient equations. Comparing with the commonly used filtered back projection (FBP) algorithm, the MAP-EM algorithm reduced noise in polyethylene region on the reconstructed images at two tube voltages with 80 kVp and 140 kVp up to 33.23%, 26.43% respectively. The contrast-to-noise ratios (CNRs) of the aluminum images were improved by 54.13% and 41.02%, that of the HA images by 52.75% and 40.47%, and that of the salt water images by 52.84% and 40.22% at the above two tube voltages. Compared with DE-FBP, the decomposition error of DE-MAP was reduced by 95.94% to 99.09%. In conclusion, the result shows superior performance on material decomposition and identification with high CNRs and low decomposition errors with the DECT images reconstructed by MAP-EM algorithm.

为了提高肺癌放疗计划危及器官勾画的精度和效率,提出了一种基于带孔U-net神经网络的肺癌放疗计划危及器官肺及心脏的并行分割方法.首先,构建了肺窗、心脏窗以及纵膈窗下的三通道伪彩色图像数据集,将图像数据集分成训练集、验证集以及测试集;然后,搭建了带孔U-net神经网络,利用训练集和验证集对其进行训练和参数调优;最后,利用测试集对训练后的带孔U-net神经网络进行图像分割性能评价,并与U-net神经网络及3种传统图像分割算法进行比较.实验结果表明,带孔U-net神经网络分割性能最优,可有效地完成肺及心脏的自动并行分割,提高勾画效率,分割结果与人工勾画结果相当.