基于集成深度学习的有源干扰智能分类

针对现有基于机器学习的雷达有源干扰分类大多需要构建人工特征集且小样本情况下分类精度低的问题, 提出一种基于多通道特征融合的集成卷积神经网络(convolutional neural network, CNN)分类方法。首先, 建立多种有源干扰的数学模型, 仿真并利用短时傅里叶变换获得其时频分布图; 其次, 提取时频分布图的实部、虚部和模值三通道特征, 通过多种特征组合方式建立不同特征组合的样本集; 最终, 构建以CNN为基分类器的集成深度学习模型, 每个CNN分别提取不同样本集的特征, 对所有基分类器的预测结果做多数投票得到集成模型的整体预测结果。实验表明, 该方法能够有效实现小样本情况下多类有源干扰的高精度智能化识别。     

Reinforcement learning-based scheduling of multi-battery energy storage system

In this paper, a reinforcement learning-based multibattery energy storage system(MBESS) scheduling policy is proposed to minimize the consumers ’ electricity cost. The MBESS scheduling problem is modeled as a Markov decision process(MDP) with unknown transition probability. However, the optimal value function is time-dependent and difficult to obtain because of the periodicity of the electricity price and residential load. Therefore, a series of time-independent action-value functions are propos...     

基于SOPC的深空目标实时跟踪系统

提出一种用可编程片上系统技术实现深空目标检测跟踪系统的方法,该系统采用单片FPGA构建双核系统,其中一个CPU进行检测跟踪,另一个CPU进行通信控制,将软硬件设计有机融为一体。根据不同算法对数据精度要求的不同,将浮点运算和定点运算结合在一起,并使用定制指令、自定义逻辑和C2H等硬件加速技术,减少了处理时间。实测结果表明,该系统可对深空红外目标进行实时检测跟踪。     

A review of mobile robot motion planning methods:from classical motion planning workflows to reinforcement learning-based architectures

Motion planning is critical to realize the autonomous operation of mobile robots. As the complexity and randomness of robot application scenarios increase, the planning capability of the classical hierarchical motion planners is challenged. With the development of machine learning, the deep reinforcement learning(DRL)-based motion planner has gradually become a research hotspot due to its several advantageous feature. The DRL-based motion planner is model-free and does not rely on the prior stru...     

Joint mission and route planning of unmanned air vehicles via a learning-based heuristic

Unmanned air vehicles(UAVs) have been regularly employed in modern wars to conduct different missions. Instead of addressing mission planning and route planning separately,this study investigates the issue of joint mission and route planning for a fleet of UAVs. The mission planning determines the configuration of weapons in UAVs and the weapons to attack targets, while the route planning determines the UAV’s visiting sequence for the targets. The problem is formulated as an integer linear progr...     

基于深度学习算法的坦克装甲目标自动检测与跟踪系统

实现对目标的自动检测与跟踪是坦克火控系统未来发展的重要方向。首先采用迁移学习的方法将基于深度学习模型的Faster R-卷积神经网络(faster R-convolution neural network,Faster R-CNN)算法应用解决复杂背景下的坦克装甲目标检测问题,与基于人工模型的传统算法相比达到了较高的检测精度。其次,针对坦克火控系统现有目标跟踪算法的不足,通过将Faster R-CNN算法与现有跟踪算法相结合,提出了复合式目标跟踪算法,实现了对坦克装甲目标的自动检测与稳定跟踪。最后,设计了一套目标自动检测与跟踪系统,采用动态扫描凝视成像技术实现了对大范围战场图像的快速、清晰获取,并对所提算法进行了实验测试。同时也指出了深度学习方法在应用于坦克火控系统时仍然存在的部分问题。     

双目视觉在粒子图像测量中的研究

针对粒子测速系统中空间点立体匹配难、空间信息难以表示的问题,构建了基于双目视觉的三维粒子测速一般模型。模型中提出并采用了分层提取的逐层相关法:即先对三维空间进行二维切分,然后结合粒子图像测速(particle image velocimetry,PIV)技术与粒子追踪测速(particle tracking velocimetry,PTV)技术对不同帧的二维粒子层面进行相关处理并求出粒子运动矢量。最后平面校正实验、空间切片重构实验以及人工合成粒子图的测速试验均证明了该方法的可靠性,使用该方法可以较精确的得到可测范围内空间任意点的矢量场。     

基于深度学习特征迁移的装备体系效能预测

针对武器装备体系效能评估在高维噪声小样本数据条件下准确性不高的问题,提出一种基于堆栈降噪自编码与支持向量回归机的混合模型。利用堆栈自编码神经网络对通用深层特征的自主抽取能力,通过在相似源域大数据上预训练混合模型,获得两任务间的共有特征知识,借助对该知识的迁移,在目标域微调该混合模型,从而提升支持向量回归机在小样本噪声数据上的学习预测精度。在一定作战想定背景下,结合武器装备体系仿真试验数据,对该混合模型进行验证。实验结果表明,与传统支持向量回归机等模型相比,所提模型能够更准确地评估装备效能。     

面向日地拉格朗日L2点探测的深空遥测数传系统设计与试验

嫦娥二号卫星首次实现从月球轨道飞向日地拉格朗日L2点开展深空探测试验。受当前星地测控设备天线口径限制,遥测数传系统面临作用距离远、接收信号弱的技术难题,为此开展了地面弱信号捕获、低损失数据解调和高效信道译码等新技术探索与试验验证。试验结果表明,地面遥测数传系统成功实现了1 500 000~1 700 000 km的数据接收解调,并具有实现更远距离深空通信的拓展能力。     

基于双目视觉的空中加油锥套定位与对接控制

针对插头-锥套式自主空中加油对接过程中实时获取加油锥套空间位置及控制导引的问题, 提出了一种结合深度学习(改进的YOLOv4-Tiny)和双目视觉匹配的快速定位方法。通过插入空间金字塔池化(spatial pyramid pooling, SPP)模块和修改部分卷积层结构, 改进后的YOLOv4-Tiny对416×416的输入检测速度达到182 Hz。与原网络相比, 体积减小20.47%, 在测试集上的平均交并比提高5%;制作了加油锥套的缩比模型进行开展视觉定位实验, 实验中平均深度预测误差小于5%, 空间位置预测符合预期。通过引入一种建立在Yolo预测基础上的快速边缘拟合方法, 获得锥套的椭圆形特征。此外, 建立了一种基于投影算子的模型参考自适应控制(model reference adaptive control, MRAC)增广控制器, 在受油机机体坐标系下跟踪锥套目标。仿真结果显示, 受油机在锥套平面中的平均跟踪误差小于加油锥套的捕获半径, 满足对接要求。     

A deep reinforcement learning method for multi-stage equipment development planning in uncertain environments

Equipment development planning(EDP) is usually a long-term process often performed in an environment with high uncertainty. The traditional multi-stage dynamic programming cannot cope with this kind of uncertainty with unpredictable situations. To deal with this problem, a multi-stage EDP model based on a deep reinforcement learning(DRL) algorithm is proposed to respond quickly to any environmental changes within a reasonable range. Firstly, the basic problem of multi-stage EDP is described, and...     

基于深度强化学习的卫星光网络波长路由算法

针对由卫星光网络拓扑动态变化、业务多样化和负载不均引起的路由收敛慢和波长利用率低的问题, 提出了一种基于深度强化学习的卫星光网络波长路由分配方法。基于软件定义中轨/低轨(medium earth orbit/low earth orbit, MEO/LEO)双层卫星网络架构, 利用深度强化学习算法动态感知网络当前的业务负载和链路状况, 构造基于时延、波长利用率和丢包率的奖励函数进行选路决策。为了解决单跳链路对整个光路的影响, 引入链路瓶颈因子, 搜索符合服务质量(quality of service, QoS)约束的最优路径。研究结果表明, 与传统卫星网络分布式路由(satellite network distributed routing algorithm, SDRA)算法和Q-routing算法相比, 所提算法降低了网络的时延、丢包率, 提高了波长利用率, 同时也降低了高优先级业务的阻塞率。     

环境风险知觉的空间折扣模型

环境风险有着典型的空间特征, 但空间距离对于风险知觉的影响机制与模式还缺乏深入研究. 该研究从环境风险现实背景中提炼出科学化问题, 提出了一个面向空间风险的偏好公理体系, 推导出指数模式的概率-空间距离权衡关系, 并用一个空间折扣模型刻画了个体的环境风险评价方式. 模型表明, 个体在心理上将远处的风险事件映射到近端位置, 而空间距离导致一种风险感知的折扣效应, 表现为个体降低了对环境后果严重性及其可能性的主观估计. 该模型首次依托一般性的环境风险事件建立了空间距离与概率之间的显性联系, 为空间背景下的风险研究提供了一个有用的框架. 该研究也有助于理解社会个体的环境风险知觉方式, 对于环境风险管理有着重要启示.     

A neuroendocrine-inspired bionic manufacturing system

Due to rapid changes of markets and pressures of competitions,the manufacturing companies are forced to adapt their production ways to support diversity of customer’s needs and increase of new product developments.As biological organisms are quite capable of adapting to environmental changes and stimulus,bio-inspired concepts have been recognized much suitable for adaptive manufacturing system control.This paper,therefore,proposes a novel concept of NeuroEndocrine-Inspired Manufacturing System(NEIMS).The proposed NEIMS control architecture is based on analogies with neuro-control and hormone-regulation principles.It has the capability to explicitly specify production control schemes including control points,material,information flow paths and logical operations,and to agilely deal with the frequent occurrence of unexpected disturbances at the shop floor level.From the cybernetics point of view,the control model of NEIMS indicates adaptive behavior to the changes in products demands due to external environment and malfunction of manufacturing cells as an internal environment.Finally,a prototype system has been set up to enable the NEIMS simulation.     

基于行为驱动的战能感知方程

作战本质上是战能积蓄感知传递转化的过程,战能感知是战能环逻辑起点,对战能环的发起和运转起着关键作用。为通过行为驱动实现装备战能与目标、环境的匹配,在总体战能约束下实现战能感知的最优化,提出了基于行为驱动的战能感知方程。首先定义了作战行为谱和战能环,明确了各对象之间的关系;其次对作战行为进行了特征解析,并进行了形式化表达,提出了作战行为谱的构建方法;然后基于行为驱动构建了战能感知方程,并用于战能感知优化控制;最后用实例对所提方法进行了验证,为作战运用优化提供技术支撑。     

HEALTHCARE： A COMPLEX SERVICE SYSTEM

Healthcare is indeed a complex service system, one requiring the technobiology approach of systems engineering to underpin its development as an integrated and adaptive system. In general, healthcare services are carried out with knowledge-intensive agents or components which work together as providers and consumers to create or co-produce value. Indeed, the engineering design of a healthcare system must recognize the fact that it is actually a complex integration of human-centered activities that is increasingly dependent on information technology and knowledge. Like any service system, healthcare can be considered to be a combination or recombination of three essential components — people (characterized by behaviors, values, knowledge, etc.), processes (characterized by collaboration, customization, etc.) and products (characterized by software, hardware, infrastructures, etc.). Thus, a healthcare system is an integrated and adaptive set of people, processes and products. It is, in essence, a system of systems which objectives are to enhance its efficiency (leading to greater interdependency) and effectiveness (leading to improved health). Integration occurs over the physical, temporal, organizational and functional dimensions, while adaptation occurs over the monitoring, feedback, cybernetic and learning dimensions. In sum, such service systems as healthcare are indeed complex, especially due to the uncertainties associated with the human-centered aspects of these systems. Moreover, the system complexities can only be dealt with methods that enhance system integration and adaptation.     

Exploring on-line societal risk perception for harmonious society measurement

The nearly 30-year economic growth miracle brings the consequent tremendous poor-rich gap leading strong drives for social transformation in current China. Chinese top leaders have realized to increase the peoples＇ income, improve quality of life and construct a ＂harmonious society＂ as key missions especially in recent 10 years. How to measure a harmonious society is one important topic as different measures may lead to different development policies. This paper outlines over 10 indices relevant to measure a harmonious society. Some are global indicators, while some are contributed by domestic researchers and arouse debates. Most of those indicators require conducting surveys on social attitudes under micro levels, which is always time consuming with problem of data quality. As Internet technology advances provide ways to record and disseminate fresh community ideas and thoughts conveniently, detecting topics or emotions from on-line public opinions is becoming a trend or one supplement way to overcome those data acquisition problems. This paper discusses one approach to on-line societal risk perception using hot search words and BBS posts. Such a trial aims to provide another way to societal risk perception different from those in traditional socio psychology studies. Challenges are also indicated.     

机制论:事物机制的系统科学分析

自然、工程、社会等各领域的事物都存在机制，人们日益重视机制问题，尤其中国社会改革开放以来“机制”成为普遍使用的词语，对于复杂问题需要从机制上解决，但至今对“机制”还缺少专门研究、缺乏专门理论。运用系统科学、复杂性理论深入研究事物机制的概念、原理、类型、作用与设计，初步建立起系统科学一门新分支理论：机制论。     

A TWO-PHASE APPROACH TO FUZZY SYSTEM IDENTIFICATION

A two-phase approach to fuzzy system identification is proposed. The first phase produces a baseline design to identify a prototype fuzzy system for a target system from a collection of input-output data pairs. It uses two easily implemented clustering techniques: the subtractive clustering method and the fuzzy c-means (FCM) clustering algorithm. The second phase (fine tuning) is executed to adjust the parameters identified in the baseline design. This phase uses the steepest descent and recursive least-squares estimation methods. The proposed approach is validated by applying it to both a function approximation type of problem and a classification type of problem. An analysis of the learning behavior of the proposed approach for the two test problems is conducted for further confirmation.     

多处理器自适应干扰抑制系统

本文描述了一个采用多处理器的自适应干扰抑制系统，讨论了系统的拓扑结构、组成框图及测试环境、自适应处理算法，给出了用于雷达数字波束形成系统自适应干扰抑制的实验结果。