基于混合神经网络的无线电信号自动调制识别

随着无线电信号数据海量增加，复杂电磁环境下面临着未知威胁和目标侦察识别复杂度高的问题，本文针对未知无线电信号的特征提取任务，设计了一种混合神经网络以提高目标无线电信号的识别能力。先通过胶囊神经网络对未知信号的空间信息进行提取，再进一步运用门控循环单元提取信号在时间上的特征信息。设计混合网络模型将信号的时间和空间特征相结合，提高对目标信号的分类精度。通过RML2016.04C调制信号数据集，验证了混合神经网络的识别性能。结果表明:当信噪比为6 dB时，混合网络模型对多种不同调制信号的分类精度大于95%。因此，本文所设计的混合神经网络能够有效对不同调制信号进行准确分类。     

Model predictive control for close-proximity maneuvering of spacecraft with adaptive convexification of collision avoidance constraints

This study investigated model predictive control (MPC) for close-proximity maneuvering of spacecraft. It is essential for a designed MPC to effectively handle collision avoidance between the servicer spacecraft and the client spacecraft, especially while the client is rotating. The rotating motion of the client leads to dynamic changes in the collision avoidance constraints, which increases the difficulty of optimizing the control input in the MPC framework. Therefore, this study presents a method to improve the performance and computational efficiency of MPC for rendezvous and docking with a nonrotating or rotating client. An ellipsoid is adopted to model the client’s keep-out zone (KOZ). Given the spherical KOZ of the servicer, an expanded ellipsoid is introduced to describe the KOZ for the center of mass of the servicer and modeled as a nonlinear constraint. The linearization method for reference points located at the expanded ellipsoid is adopted to convexify the nonlinear constraints. The reference points are adaptively determined according to the positions of the servicer, client, and expanded ellipsoidal KOZ. The resulting hyperplanes are then used to describe the collision avoidance constraints. By utilizing the aforementioned strategies, combined with the calculated reference points, an adaptive convex programming algorithm suitable for real-time implementation of MPC is derived. The performance of the proposed method is demonstrated through numerical simulations.     

基于可穿戴惯性测量的滑雪运动员姿态测量与水平评估方法

运动量化分析是科学化滑雪训练的重要发展方向,而滑雪过程中人体运动模式的数字化表达是科学量化分析的基础和依据.设计并搭建了基于可穿戴MEMS惯性测量单元的人体运动捕捉与姿态重构系统,结合人体多刚体运动模型实现了滑雪过程的人体重构,并应用于高山滑雪运动中回转动作的辅助训练.同时,针对基于多自由度模拟滑雪训练平台的室内训练场景,提出了一种适用于滑雪回转运动的数字化评估方法.该方法利用动作捕捉系统和姿态重构系统提取滑行者五种滑行特征数据,并通过与高水平运动员运动参数的相似性度量和线性拟合实现对滑雪运动员技术参数的评估,以辅助科学训练.最后在室内Olymp模拟滑雪训练台上进行实验,验证了该方法的有效性.     

载人月球科研试验站建设规划及设想

进入21世纪以来，月球再次成为太空探索的焦点，许多国家提出各自的探月计划。区别于在美国60年前完成的阿波罗工程，未来月球探测的重点聚焦月球的长期建设与开发。如何充分利用当前的技术积累，在有限的运载能力和合理的发射运营成本条件下，建设载人月球科研试验站，实现月球极端环境下的长期运行是一个值得关注和研究的课题。结合当前国内外载人航天与探月工程发展情况，提出我国载人月球科研试验站人货分落、分段建设的规划设想。通过对安全性设计、乘组规模、环控生保需求、选址等展开分析，明确初步月球科研试验站建设需求，并提出满足2人长期月面驻留“2+1”构型月球科研试验站方案。     

基于改进的卷积神经网络的步态识别

为了提高惯性传感器采集到的序列数据中步态识别的准确率，建立了一个激励层改进的卷积神经网络(CNN)模型。针对三轴加速度传感器对运动太过敏感导致步态周期划分不准确的问题，采用加速度传感器与弯曲度传感器组合获取人体运动信息。将CNN模型中激励层的线性整流函数（ReLU）改进为带泄露线性整流函数(Leaky ReLU)，以解决遇到卷积输出数据小于0时神经元被抑制的问题，进而达到提高步态识别准确率的目的。实验结果表明：激励层优化的CNN模型在行走、上下楼和上下坡五种步态模式下识别率达到了95.79%，与未采用弯曲度传感器的改进CNN模型和未进行激励层改进的CNN模型相比，步态识别率有所提高。     

基于TestBed 和Visual Studio 对QT 程序单元 测试的研究与实现

针对测试工具TestBed 无法动态执行QT 程序单元测试用例的问题，文章通过介绍单元测试定义、测 试原理以及TestBed 工具的使用原理、并指出了TestBed 工具对QT 程序进行单元测试存在的问题，最终提出 了一种结合TestBed 和Visual Studio 对QT 程序进行单元测试的方法，并给出了方法的详细设计与实现步骤， 为QT 程序动态执行单元测试提供一种简单、便捷的实现方式。实际结果表明，结合TestBed 与Visual Studio 对QT 程序的单元测试方法显著提高了单元测试的效率，减少了单元测试的成本。     

Quick identification of guidance law for an incoming missile using multiple-model mechanism

A guidance law parameter identification model based on Gated Recurrent Unit (GRU) neural network is established. The scenario of the model is that an incoming missile (called missile) attacks a target aircraft (called aircraft) using Proportional Navigation (PN) guidance law. The parameter identification is viewed as a regression problem in this paper rather than a classification problem, which means the assumption that the parameter is in a finite set of possible results is discarded. To increase the training speed of the neural network and obtain the nonlinear mapping relationship between kinematic information and the guidance law parameter of the incoming missile, an output processing method called Multiple-Model Mechanism (MMM) is proposed. Compared with a conventional GRU neural network, the model established in this paper can deal with data of any length through an encoding layer in front of the input layer. The effectiveness of the proposed Multiple-Model Mechanism and the performance of the guidance law parameter identification model are demonstrated using numerical simulation.     

某船载无人操控天线末段跟踪的数引策略改进

从无人操控天线需要依赖数引完成目标自动捕获以及现有的主动段航天测量的数引模型精度不够的背景出发,指出提高数引精度即进行数引改进的必要性。然后结合末段跟踪的特殊性,从时差、视差和延时累积误差三个影响数引精度的最主要方面详细分析了它们的形成机理,并给出了各自的改进方法。