具有热阻、热漏和内不可逆性的联合卡诺热机生态学优化

用有限时间热力学的方法分析具有热阻、热漏、内不可逆性的定常流联合卡诺型热机循环.导出了在傅立叶导热定律下联合循环功率、效率和生态学指标的性能,并进行优化;得到功率、效率和生态学指标之间的优化关系,并由数值计算分析了功率、效率和循环熵产率之间的关系.所得的结果表明,最大生态学指标下的效率十分接近于联合循环可以达到的最大效率;相应的熵产率也要低于以输出功率为优化目标时的熵产率.     

基于退化速率跟踪粒子滤波的剩余使用寿命预测框架

毋庸置疑,剩余使用寿命预测对于设备的健康管理越来越重要。近年来粒子滤波方法被越来越多地应用到设备寿命预测技术当中,这是因为粒子滤波方法能更好的解决非线性非高斯系统滤波问题,而且能够获得不确定度信息。但该方法的预测性能却过度依赖于预测模型,并且对于模型参数的初始分布也比较敏感,这在一定程度上限制了粒子滤波预测方法的进一步发展。本文针对基本粒子滤波预测方法的不足,提出了一种基于退化速率跟踪粒子滤波的通用预测框架,以历史观测数据的退化速率统计规律作为指导来跟踪目标数据的退化速率,实现对粒子滤波预测方法的简化。并将该方法用于轴承和锂离子电池的剩余使用寿命预测,验证了方法的有效性。     

NONPROLIFERATION POLICY CROSSROADS:

On October 1, 2008, Congress enacted a proposal that originated with President George W. Bush in 2005 to approve an unprecedented nuclear trade pact with India by removing a central pillar of US nonproliferation policy. Despite the numerous political challenges confronting the Bush administration, the initiative won strong bipartisan support, including votes from Democratic Senators Joseph Biden, Hillary Clinton, and Barack Obama. The four-year struggle to pass the controversial US-India nuclear trade agreement offers an exceptionally valuable case study. It demonstrates a classic tradeoff between the pursuit of broad multilateral goals such as nuclear nonproliferation and advancement of a specific bilateral relationship. It reveals enduring fault lines in executive branch relations with Congress. It vividly portrays challenges confronting proponents of a strong nonproliferation regime. This article is based on an analysis of the negotiating record and congressional deliberations, including interviews with key participants. It assesses the lessons learned and focuses on three principal questions: how did the agreement seek to advance US national security interests?; what were the essential elements of the prolonged state-of-the-art lobbying campaign to win approval from skeptics in Congress?; and what are the agreement's actual benefits—and costs—to future US nonproliferation efforts?     

Estimating non‐transparent military expenditures: The case of china (PRC)∗

Data limitations arising from non‐transparency in military expenditure reporting impose significant constraints on economists studying country and comparative developments in this field. This article focuses on the case of China. It reviews the recent literature on China's military expenditure, examines the variance in published estimates, and identifies three factors which help explain the variance: budgetary falsification; non‐budgeted PLA revenue; and the US dollar purchasing‐power parity of the yuan. Conclusions emphasize important policy implications of such variance and suggest a more active role for defence economists in a field of research generating considerable attention from international relations scholars.     

RELIABLE ENERGY SUPPLY AND NONPROLIFERATION

The phenomenon of global warming has led to a revival of the prospects for increased nuclear energy production worldwide, yet such increased production carries with it the increased risk of proliferation. To mitigate this risk, various multinational arrangements have been proposed to provide reliable supply of nuclear fuel while at the same time discouraging the construction of national plants for nuclear enrichment and reprocessing. This article provides a brief history of some of these proposals and concludes that the likelihood of success for such schemes as effective tools for nonproliferation is not high at this time. A proposal from the World Council on Renewable Energy to expand the understanding of supplier obligations under Article IV of the Treaty on the Non-Proliferation of Nuclear Weapons (NPT) to include the development of non-nuclear energy technologies for NPT parties in good standing is potentially a much better nonproliferation tool. Such an approach tracks the ideas contained in Title V of the Nuclear Non-Proliferation Act of 1978, which has recently received revived congressional interest.     

LDO稳压器敏感度建模与仿真技术研究

基于LDO稳压器在电磁干扰（EMI）下产生直流偏移失效的机理分析,展开敏感度建模与仿真方法研究。使用一款实验芯片,创新地引入片上电压传感器,用于测试EMI在LDO稳压器内部的传播特性。在敏感度建模中,建立等效电路模型,通过直流功能测试,Z参数阻抗特性测试验证模型的正确性,将该模型用于LDO稳压器的敏感度预测。在敏感度仿真过程中,通过分析关键子电路和不断增加寄生元件,仿真不同寄生因素对敏感度影响的权重。将仿真结果与传导直接注入法（DPI）片上测试结果对比,仿真结果与DPI测试在频域1MHz至1GHz匹配。     

基于CPU-GPU混合计算平台的RNA二级结构预测算法并行化研究

RNA二级结构预测是生物信息学领域重要的研究方向,基于最小自由能模型的Zuker算法是目前该领域最典型使用最广泛的算法之一。本文基于CPU GPU的混合计算平台实现了对Zuker算法的并行和加速。根据CPU和GPU计算性能的差异,通过合理的任务分配策略,实现二者之间的并行协作计算和处理单元间的负载平衡;针对CPU和GPU的不同硬件特性,对Zuker算法在CPU和GPU上的实现分别采取了不同的并行优化方法,提高了混合加速系统的计算性能。实验结果表明,CPU处理单元在混合系统中承担了14%以上的计算任务,与传统的多核CPU并行方案相比,采用混合并行加速方法可获得15.93的全局加速比;与最优的单纯GPU加速方案相比,可获得16%的性能提升,并且该混合计算方案可用于对其它生物信息学序列分析应用的并行和加速。     

融合可修性度量和双线学习的反导预案修正

对反导预案应用中基于预案推理的预案修正技术进行了研究,提出了一种新的反导预案修正方法。首先,在定义预案属性否决权和修正裕度的基础上,设计了一种考虑预案可修性的相似度计算方法;然后,以其为预案检索和修正规则学习的导向,融合离线预案库学习和在线实时学习策略,给出了预案修正规则的双线学习方法;建立了融合预案可修性度量和双线学习的反导预案修正模型,并对规则的归纳学习策略进行了研究;最后,进行了实例验证,结果证明了新修正方法的可行性和有效性。     

数据与知识双驱动的备件需求模糊预测模型

针对知识驱动型需求预测模型所需的专家知识稀缺、数据驱动型需求预测模型可解释性不足的问题,提出了数据与知识双驱动的备件需求模糊预测模型。该模型基于模糊聚类算法将数值型数据聚类为结构简单、可解释性强的规则库,运用模糊逻辑将领域专家知识表示为Mamdani型规则库。在此基础上,引入了一种新型智能计算理论——模糊网络理论对两类规则库进行合并运算,形成初始预测模型。采用遗传算法优化模型规则库的模糊集参数来提高模型预测准确性。通过与模糊聚类算法进行对比,提出的模型在可解释性以及准确性指标上均具有优势。     

面向混部云失败批处理作业的预测算法

为了降低混部云失败批处理作业的风险,使用K-means聚类算法将批处理作业分为四类,在分类的基础上提出了二层嵌套分类模型(two-layer nested classification model, TLNM),实现了基于TLNM的预测算法。基于Ali Trace 2018数据集上的实验结果表明,该算法的接受者操作特性(receiver operating characteristic, ROC)曲线明显优于其他常用分类器,ROC曲线下面积(即AUC)可以达到0.978,表明该算法具有良好的分类性能。同时召回率可以达到0.951,通过混淆矩阵可以看出TLNM算法能够准确预测出执行失败的批处理作业。