图象存储体与PC机接口设计

本文研究利用PC机扩充槽设计图象存储体与PC机的DMA接口,包括接口电路和接口驱动程序。主要完成图象存储体与PC机内存间DMA数据传送的启动及相应控制信号的产生、转换,并提供数据通道,本接口应用于“脉冲X光医疗诊断系统”,对其它图象处理系统的计算机接口亦有参考价值。     

顺序PROLOG机KD-PP的系统结构和硬件实现技术

本文提出的KD-PP系统是一种基于编译技术的顺序PROLOG推理系统,该系统的设计为逻辑型程序语言PROLOG的实现提供了硬件支持,因而能高效地执行PROLOG程序。本文从数据表示、存储系统、机器状态和指令系统等方面全面地介绍了顺序PROLOG机KD-PP的系统结构和硬件实现技术。     

基于DRFM的机载PD雷达干扰研究

针对传统干扰技术对PD雷达干扰的不足,噪声干扰对干扰机功率要求高,欺骗干扰产生的干扰数目单一,基于当前对抗先进体制雷达的手段——DRFM,对PD雷达干扰进行研究。通过将截获的雷达信号存储、处理,按程序设定的转发间隔重复读出当前采样信号并转发,产生了多个距离欺骗假目标;通过对每次转发干扰进行频率调制,产生了包含距离速度欺骗的一系列假目标;并经过Matable仿真验证了在PD雷达窄带滤波器中,出现多个距离或速度欺骗假目标。并通过仿真验证了干扰的有效性。     

基于磁记忆检测技术的即时报警功能设计与实现

根据金属磁记忆检测技术中缺陷判定依据的运算特点,为减少程序计算量、提高缺陷检测效率,改进了传统缺陷判定算法．在检测设备基本结构基础上增加了信号调理卡,并借助多功能采集卡的数字输出功能实现了缺陷检出时的即时报警,大幅提高了检测系统的工作效率．     

基于小波包能量谱的管道缺陷磁记忆检测信号特征研究

现有磁记忆检测技术判定准则,只能指示应力集中位置,无法进一步获取应力集中信息。为获取应力集中信息,提出一种基于小波包能量谱的磁记忆信号分析方法,进行试件拉伸试验。拉伸应力为200 MPa时,信号小波包能量谱分布较为均匀,各频带能量占总能量之比均小于15%,不存在集中分布的频带范围。拉伸应力为410 MPa时,信号小波包能量谱最大值分布在1,3,4频带,1~4频带能量之和占总能量的73.8%,小波包能量主要集中在低频段。试件屈服后,信号小波包能量谱最大值分布在1,2频带,能量谱分布极为分散,能量主要集中在低频段的1,2频带,1~3频带能量之和占总能量的87.3%。管道试件应力集中程度与磁记忆信号的小波包能量谱分布特征有关,应力集中程度越低,小波包能量谱分布越均匀;应力集中程度越高,小波包能量谱分布越集中,能量主要向低频段集中。     

Russian Reactions to Reagan's ‘Maritime Strategy’

The US Navy's audacious Maritime Strategy of the 1980s is often credited with acting as a key catalyst to the demise of the USSR. In assessing the role of strategic missile submarine “bastions” in the country's overall military strategy, the authors are skeptical of the above thesis, explaining that Moscow actually viewed deployments of Tomahawk land-attack cruise missiles on to US submarines (versus direct threats against the Soviet Navy itself) as the most alarming threat. The authors concede that the arms race played a role in the collapse of the USSR, but deny a direct connection with Soviet naval development citing the relatively minor naval proportion of overall defense spending. For the difficulties facing the fleet at the end of the Cold War, they instead blame deep systemic problems internal to the Soviet Navy.     

The terrorists of today are the heroes of tomorrow: the anti-British and anti-American insurgencies in Iraqi history

The anti-British rebellion of 1920 has become one of the most important foundational myths of Iraqi nationalism. It has been the subject of poetry, theatre, film and folklore and continues to be commemorated and evoked as a symbol of the Iraqi nation state and of Iraqi pride and independence. This popular reading of events ignores the realities of the rebellion: that it was a mid-Euphrates, rather than a national, affair and that the motivations, to begin with at least, were far more personal and economic than national. Nevertheless, the rebellion's memory has acquired a hallowed place in Iraqi nationalist discourse. This article will focus on two points: first, to place the rebellion of 1920 in its correct historical context; second, it will be argued that today's anti-American insurgency will make the same historiographical journey from being a tumultuous and localised event that had no shortage of detractors to becoming a defining symbol of Iraqi nationalism. As will be shown, there is already much evidence to suggest that the process has already begun. The foreign troop presence cannot remain indefinitely and when, sooner or later, Coalition forces withdraw we may well see the development of a narrative that credits the ‘resistance’ for gaining Iraqi independence; in other words, the ‘resistance’ today may well become the foundational nationalist myth of the new Iraq.     

基于记忆有理函数的功率放大器行为模型

针对功率放大器的非线性特性及记忆效应,提出了一种基于记忆有理函数的功放行为模型。在传统记忆多项式模型和无记忆有理函数模型基础上,构建记忆有理函数模型,并利用共轭梯度法辨识模型系数,同时比较不同记忆深度和不同非线性阶数下的归一化均方误差,获取最佳记忆深度和非线性阶数。采用多载波的WCDMA信号和MRF6S21140H功放来验证模型的有效性,并与记忆多项式模型、无记忆有理函数模型进行了比较。结果表明,记忆有理函数模型在减少系数数目的同时具有更好的逼近精度。     

利用循环网络的降落伞开伞载荷补偿方法

为了对降落伞充气展开过程中的开伞载荷进行更加准确的预测,提出一种基于循环神经网络的开伞载荷补偿计算方法,包括模型架构和数据处理方式。该方法将充气时间法计算的预测值代入循环网络进行二次计算,使最终结果能够更加贴近试验真值。使用多层前馈网络、标准循环网络与长短时记忆网络三种网络进行比较,验证了所提模型预测结果的适用性和准确性,研究了学习率、输入层维度和隐层维度等超参数对模型性能的影响,并给出了基于长短时记忆网络的补偿模型最优训练条件。实验结果表明,利用循环网络进行开伞载荷预测具有较好的拟合结果,为机器学习与降落伞工业的学科交叉研究提供了一定的参考方向。     

Maintenance scheduling for modular systems: Modeling and algorithms

We study new models of scheduled maintenance management for modular systems, consisting of multiple components with respective cycle limits. The cycle limit of each component specifies the time interval in which this component must be repaired or replaced. The goal is to compute a feasible maintenance schedule that minimizes the cost associated with component maintenance. Applications of these models arise in Air Force aircraft maintenance as well as in other arenas with required preventive maintenance. The typical cost structures that arise in practical settings are submodular, which make the resulting models computationally challenging. We develop two efficient and operationally tenable approximation algorithms. We prove constant factor worst‐case guarantees for both algorithms, and present computational experiments showing that these algorithms perform within a few percent of optimality on operationally relevant instances. © 2014 Wiley Periodicals, Inc. Naval Research Logistics 61: 472–488, 2014