基于函数链神经网络的深度分类器

目前的宽度学习系统（Broad learning system，BLS）通过所建立的一系列映射节点和增强节点来形成联合节点。因为联合节点与输出层的线性连接，网络权值可以用求解伪逆的方法快速求得，避免了耗时的训练过程，从而成为快速而高效的学习方法。然而在追求高精度结果的过程中，BLS对于增强节点数量的需求过于巨大，容易造成过拟合问题。为此，本文提出了基于函数链神经网络（Functional-link neural network，FLNN）的深度分类器（FLNN based deep classifier，FLNNDC），旨在提供一种更加简单却又不失精度的BLS变体结构。FLNNDC将几个轻量级的BLS子系统堆积成栈式结构，每一个轻量级的BLS子系统随机选择一部分映射节点生成增强节点，而不是全部映射节点。和原宽度结构相比，在几个主流数据集上的实验结果表明本文所提出的FLNNDC分类器具有网络结构更小且学习速度更快的优势。     

六自由度激励台精密高刚度铰链结构设计与试验应用

精密高刚度铰链作为六自由度并联激励平台中的关键组件,其性能高低对激励台的控制精度有直接影响。针对铰链所需满足的高刚度、高基频和无间隙等要求,文章提出了基于虎克铰结构的铰链设计方法。首先根据铰链所处的工况确定选用圆锥滚子轴承作为旋转支撑部件,并通过对轴承的配置和预紧实现无间隙振动运动的要求,从而确定了铰链结构形式。然后,利用有限元法校核了结构的刚度与基频,进而完成铰链结构方案和样机制造,并将样机集成于激励台中。最后,开展了铰链动态特性试验和激励台振动控制试验,分别获取了铰链的基频和激励台控制响应。结果表明,铰链可满足激励台工作频段内的使用要求,铰链设计合理。     

An improved numerical method for constructing Halo/Lissajous orbits in a full solar system model

An improved numerical method that can construct Halo/Lissajous orbits in the vicinity of collinear libration points in a full solar system model is investigated. A full solar system gravitational model in the geocentric rotating coordinate system with a clear presentation of the angular velocity relative to the inertial coordinate system is proposed. An alternative way to determine patch points in the multiple shooting method is provided based on a dynamical analysis with Poincare′sections. By employing the new patch points and sequential quadratic programming, Halo orbits for L_1, L_2, and L_3 points as well as Lissajous orbits for L_1 and L_2 points in the EarthMoon system are generated with the proposed full solar system gravitational model to verify the effectiveness of the proposed method.     

Fault mode probability factor based fault-tolerant control for dissimilar redundant actuation system

This paper presents a Fault Mode Probability Factor (FMPF) based Fault-Tolerant Control (FTC) strategy for multiple faults of Dissimilar Redundant Actuation System (DRAS) composed of Hydraulic Actuator (HA) and Electro-Hydrostatic Actuator (EHA). The long-term service and severe working conditions can result in multiple gradual faults which can ultimately degrade the system performance, resulting in the system model drift into the fault state characterized with parameter uncertainty. The paper proposes to address this problem by using the historical statistics of the multiple gradual faults and the proposed FMPF to amend the system model with parameter uncertainty. To balance the system model precision and computation time, a Moving Window (MW) method is used to determine the applied historical statistics. The FMPF based FTC strategy is developed for the amended system model where the system estimation and Linear Quadratic Regulator (LQR) are updated at the end of system sampling period. The simulations of DRAS system subjected to multiple faults have been performed and the results indicate the effectiveness of the proposed approach.     

A holistic approach to SIM platform and its application to early-warning satellite system

This study proposes a new simulation platform named Simulation Integrated Management (SIM) for the analysis of parallel and distributed systems. The platform eases the process of designing and testing both applications and architectures. The main characteristics of SIM are flexibility, scalability, and expandability. To improve the efficiency of project development, new models of early-warning satellite system were designed based on the SIM platform. Finally, through a series of experiments, the correctness of SIM platform and the aforementioned early-warning satellite models was validated, and the systematical analyses for the orbital determination precision of the ballistic missile during its entire flight process were presented, as well as the deviation of the launch/landing point. Furthermore, the causes of deviation and prevention methods will be fully explained. The simulation platform and the models will lay the foundations for further validations of autonomy technology in space attack-defense architecture research.     

不同密度弹丸对水冰的超高速撞击成坑实验

为了研究冰冻天体表面撞击坑的形成与演化，开展了水冰的超高速撞击成坑实验。使用二级轻气炮发射1.0 mm直径的球形弹丸，以3 km/s、5 km/s和7 km/s速度对圆柱状冰块进行撞击。弹丸材料包括聚碳酸酯和不锈钢两种，冰块温度为253 K。实验观察到了不同弹丸和不同速度条件下，冰块中撞击坑的形貌特征。对撞击坑直径、深度和剖面形状进行了测量，并与文献中铝弹丸对水冰的撞击坑进行了比较分析。获得了水冰撞击坑特征随撞击参数的变化规律，结果表明：撞击坑直径和深度的主导机制不同，坑深主要由弹丸侵彻作用形成，而坑径主要由冰块的剥落所致；坑深比坑径具有更强的对于弹丸密度的依赖性，高密度弹丸撞击坑直径具有比低密度弹丸更强的对于撞击速度的依赖性；撞击坑体积与撞击能量成正比，高密度弹丸形成的撞击坑直径表现出“能量缩比”行为，而低密度弹丸形成的撞击坑直径表现出“动量缩比”行为。     

Manifold structure preservative for hyperspectral target detection

A nonparametric method termed as manifold structure preservative (MSP) is proposed in this paper for hyperspectral target detection. MSP transforms the feature space of data to maximize the separation between target and background signals. Moreover, it minimizes the reconstruction error of targets and preserves the topological structure of data in the projected feature space. MSP does not need to consider any distribution for target and background data. So, it can achieve accurate results in real scenarios due to avoiding unreliable assumptions. The proposed MSP detector is compared to several popular detectors and the experiments on a synthetic data and two real hyperspectral images indicate the superior ability of it in target detection.     

热防护系统损伤容限分析

建立了含初始矩形损伤的热防护系统(Thermal protection system,TPS)气动热分析的CFD数值模型,分析结果表明损伤区域侧壁出现了很高的热流密度峰值,并且迎风面侧壁峰值高于背风面,而损伤区域底部热流密度却很低。利用分析获得的热流密度建立了含损伤和无损伤TPS的有限元传热分析模型。分析结果表明:损伤的存在导致防热瓦最高温度急剧上升,超过其材料能承受的极限温度(1 500℃),防热瓦首先失效,而损伤对机体最高温度影响较小。最后进行了TPS损伤容限分析,在防热瓦极限温度约束下,外部热流密度最大值从100kW/m2增加到140kW/m2,矩形损伤宽度最大容许值从22.7mm减小到12.6mm,而弧形损伤宽度最大容许值从34.6mm减小到25.1mm,即随着外部热流密度最大值增加,损伤宽度的最大容许值降低,并且相同外部热流密度水平下弧形损伤宽度的最大容许值大于矩形损伤。     

超宽带变频系统中的本振产生技术研究

本振源是超宽带变频系统中的核心部件，本文以超宽带上变频系统为例，介绍变频系统中本振信号的产生技术。本文所设计的超宽带变频系统，经过三次上变频，可将(0.8~2.8)GHz 的低频信号变换至（0.5~40）GHz。变频过程中需要一系列点频本振和一组宽带本振，基于PLL+DFS 的方法产生点频本振，基于DDS+DFS 的方式产生宽带本振。     

A novel framework for liquid propellant engine’s cooling system design by sensitivity analysis based on RSM and multi-objective optimization using PSO

One of the challenges of combustion chamber and nozzle design in a Liquid Propellant Engine (LPE) is to predict the behavior and performance of the cooling system. Therefore, while designing, the optimization of the cooling system is always of great importance. This paper presents the multi-objective optimization of the LPE’s cooling system. To this end, a novel framework has been developed, resulting from the application of the Response Surface Method (RSM) and the correlation coefficients matrix, sensitivity analysis and the The Particle Swarm Optimization (PSO). based on this method, the input variables, constraints, objective functions, and their surfaces were identified. In terms of multi-optimization algorithms, RSM and PSO are utilized to get global optimum. In conclusion, the methodology capability is to optimize the LPE’s cooling system, 6 percentage increase in total heat transfer and 7 bar decrease cooling system pressure loss, which resulted in a 1.2-seconds increase in the specific impulse of the engine.