基于迭代学习观测器的卫星姿态滑模容错控制

为解决卫星上反作用飞轮存在安装偏差、故障及外部干扰情况下的姿态控制问题,提出了一种基于迭代学习观测器的姿态容错控制方法。该方法通过设计迭代学习观测器,以较小的计算量实现对执行机构发生的故障以及安装偏差进行精确的估计。并利用观测器的观测结果设计滑模控制器,使处于外部干扰条件下的卫星系统在执行机构发生故障的情况下可以快速稳定地完成姿态机动任务。进一步基于Lyapunov稳定性定理证明了迭代学习观测器及控制器的全局渐近稳定性。针对反作用飞轮存在不确定性及故障的情况下进行仿真,仿真结果表明,与同类容错控制方法相比,所提方法可以更加快速精确地对故障进行估计并完成姿态控制。     

The modern naval fortress: an additional sea denial option for coastal states

As fixed facilities, naval fortresses seem unlikely to be important in a sea denial strategy which is usually about mobility, but new defence technologies and the changing geostrategic environment may revive the concept of the fortress. Extended ranges of anti-ship means allow onshore firepower to engage enemies over distance, even beyond the economic exclusive zones where most maritime territorial disputes occur. In the face of size limits on missile warheads that constrain their destruction of hardened targets, various active and passive defence technologies against missiles can enhance the survivability of onshore fortresses. Furthermore, onshore locations give fortresses the advantage of being unsinkable and able to accommodate greater energy and firepower capacity in contrast to vessels, as well as other mobile platforms. The onshore nature of fortresses also gives a different political meaning to being attacked, for the clear violation of sovereignty, as opposed to vessels and aircraft in a disputed space. However, the fact those fortresses are not invincible means cooperation with other existing capabilities still necessary. The case of Vietnam demonstrates how fortresses could strengthen the inferior defence capability of a coastal state vis-à-vis. a stronger sea power.     

The crisis of American military assistance: strategic dithering and Fabergé Egg armies

ABSTRACT

The United States has provided substantial amounts of military assistance and aid since the end of World War Two. During the Cold War, it proved vital in protecting numerous regimes from communist takeovers. Successful outcomes occurred when American leaders made large initial aid commitments, and the states had the capacity and political willpower to use it effectively. However, Vietnam was an example of how U.S. support lagged, as leaders in Saigon preferred political survival instead of creating regime legitimacy. Following 9/11, American security aid focused on making weak countries develop stronger security forces. Unfortunately, this created Fabergé egg militaries: expensive and easily broken by insurgents. This article suggests long-term strategic commitments need to be made alongside more resources for the American State Department and similar organizations to focus on the politics of state-building. Finally, this article suggests strategies, such as “whole-of-government” approaches, to improve long-term security and political institution building.     

基于多级RBF神经网络集成的传感器故障诊断研究

针对诊断传感器偏置故障及漂移故障的难点问题,提出了一种基于多级RBF神经网络集成的传感器故障诊断方法。该方法充分利用控制系统闭环回路测控信息,建立多级神经网络集成观测器模型。将输出与传感器实际输出相比较获取残差序列,获得基于残差序列的传感器偏置故障和漂移故障的辨识策略,实现控制系统传感器故障在线诊断。将三容水箱液位控制系统作为仿真对象,仿真结果表明该方法不仅可以提高单一神经网络的运算精度,而且采用RBF神经网络集成方式还要优于其他集成方式,可以快速准确地检测和分离传感器故障,辨识传感器故障类型、故障大小以及故障发生的时间。     

全局仿射变换条件下图像不变量提取新方法

从仿射几何理论出发,提出了一种全局仿射变换条件下图像特征不变量构造的新方法.利用仿射区域划分获取一组扩展质心集合作为图像特征点集;求解特征点集的凸包,根据扩展质心的性质将凸包顶点均匀组合成若干特征区域;利用仿射几何性质构造了有效的仿射不变量.实验对该方法进行了全面的比较分析,以验证所构造的不变量更具稳定性和高时效性.     

纯方位TMA的变增益扩展卡尔曼滤波算法

通过对水下纯方位目标运动分析中的扩展卡尔曼滤波(EKF)算法的分析,利用方位新息,构成了一条反馈回路,从而将EKF算法改进为变增益扩展卡尔曼滤波(VGEKF)算法.对比仿真分析表明,VGEKF较之EKF滤波效果有所改善,增强了稳定性,提高了精度,为水下纯方位目标运动分析的实现提供新的途径.     

一种基于观测器的自修复飞行控制系统设计

故障检测与隔离(FDI)、控制律重构是自修复飞行控制系统设计的重要内容.在自修复飞行控制系统设计中,利用观测器,设计了针对传感器的故障进行检测的方法,可使检测效果在一定的条件下,对系统参数变化不敏感,具有较好的鲁棒性;此外,当飞机操纵机构发生故障时,利用观测器,设计了基于线性模型跟随控制进行飞行重构控制的方法,使得当状态变量不能测量时的控制系统得以物理实现.最后,给出了仿真实例,结果表明了该方法的可行性和有效性.     

基于扩张观测器的三维有限时间收敛导引律

为了保证视线角速率在弹目碰撞前收敛到零附近的较小邻域内从而达到准平行接近的状态,本文基于自抗扰控制的不确定性估计补偿思想,应用反演控制方法设计了一种考虑导弹自动驾驶仪二阶动态特性和目标机动的三维有限时间收敛导引律。根据有限时间收敛控制理论,严格证明了系统的有限时间收敛特性;为抑制量测噪声,将传统跟踪微分器进行改进并应用于扩张状态观测器与反演控制的设计中。仿真结果表明:在自动驾驶仪响应延迟情况下,所设计的导引律能够导引导弹在有限时间内精确地拦截高速机动目标;改进的跟踪微分器精度高、响应快;基于改进跟踪微分器的扩张观测器估计效果理想。     

目标自动跟踪参数辨识模型的工程化动态辨识

为解决机动目标参数辨识模型工程应用问题,采用扩张状态观测器等先进的非线性滤波算法,从实测的含有观测噪声的目标位置数据的序列中,直接估计出机动目标运动的一阶、二阶、三阶导数,并结合目标运动模型辨识的理论算法,在工程应用上首次实现了机动目标运动模态和运动模型的双重动态辨识,完成了目标模型动态辨识的工程化设计.     

动态威胁环境下单无人机测向定位航迹优化算法

为解决单架无人机在动态战场环境下的测向定位问题,提出了一种基于动态窗口法的单机测向定位航迹优化算法。以最大化Fisher信息矩阵行列式为测向定位评价准则,在由动态探测雷达和静/动障碍构成的动态战场环境中,基于动态窗口法思想,将测向定位航迹优化评价准则由传统的单步最优原则扩展到对多步预测航迹的评价,同时考虑雷达探测和静/动障碍环境对预测航迹的影响,通过滚动时域方法控制无人机最优航向。仿真结果表明,所提方法能够使无人机在有效逃避雷达探测威胁以及规避环境中静/动障碍的条件下保证对目标的高精度测向定位,为解决动态战场环境下的单架无人机测向定位问题提供了新思路。