多核贝叶斯优化的模型决策树算法

构造模型决策树时超参数较多,参数组合复杂,利用网格搜索等调参方法将会消耗大量的时间,影响模型性能的提升。提出了一种多核贝叶斯优化的模型决策树算法,该算法为应对不同分类数据特性,采用三种高斯过程建模寻优,利用贝叶斯优化技术,选出最优的参数组合。实验结果表明,所提算法在参数寻优上要优于传统的模型决策树寻优方法,并且能够在迭代次数不多的情况下找到全局最优参数值,在一定程度上提升了算法的分类性能,节省了大量的调参时间。     

融合空间偏好和语义的个体活动识别方法

个体活动识别对用户画像、个性化推荐、异常行为检测、群体行为分析和基于活动的资源配置优化具有重要价值。提出了一种基于稀疏的社交媒体签到数据的个体活动语义识别方法,从签到数据中提取活动行为的时间周期性和趋势性特征,并采用空间偏好量化算法,从个体与群体活动的空间关联中提取群体和个体的空间访问偏好,使用自然语言嵌入工具BERT模型提取访问兴趣点的语义。时间特征、空间偏好特征和访问兴趣点名称语义特征共同构成表征群体、个体偏好的时空联合特征,通过极限梯度提升分类器对其进行分类,得到活动语义识别结果。在Foursquare数据集上的对比实验和消融实验中验证了所提活动语义识别模型可以有效提升活动语义识别的准确性。     

分布式异构集群中节点优先级调优算法

节点优先级常用于评价异构集群中节点的性能,因此节点优先级评价指标权重的选择非常重要。采用层次分析法(analytic hierarchy process, AHP)建立了节点优先级评价指标体系,计算得到各指标的初始权重,并使用BP神经网络对初始权重进行优化。训练时,BP网络输入为集群运行中采集的节点实时资源数据,输出为节点的优先级。分析网络训练完成后得到的权重矩阵可以获得各优先级评价指标的优化权重。实验表明,基于AHP和BP的节点优先级评价模型可以更加准确地分析节点性能。相比于Spark默认算法和权重未优化的对照算法,使用调优后的节点优先级可以有效提高集群性能。运行不同工作量的相同负载时,集群平均性能分别提高了16.64%和9.76%;处理相同工作量的不同负载时,集群的平均性能分别提高了12.49%和6.54%。     

基于本体技术的分布式模拟训练系统操作监控平台

针对分布式模拟训练系统操作训练特点,在基于HLA（High Level Architecture）的模拟训练系统中,提出并建立了基于本体技术的操作训练监控平台,解决了训练系统和监控平台中操作消息的语义异构问题,实现了操作训练情况的动态实时显示。在平台实现中,采用RTI（Run-Time Infrastructure）提供的服务实现了联邦数据收发;建立了操作信息领域本体,采用计算语义相似度的方法实现操作信息由分布交互层向语义层映射;通过专家知识库和操作监控引擎判断操作逻辑合理性;利用View Graph以图形化的方式展示出操作训练情况。最后,通过一个炮车操作训练监控实例,验证了该平台的可行性和有效性。     

Dynamic analysis of a guided projectile during engraving process

The reliability of the electronic components inside a guided projectile is highly affected by the launch dynamics of guided projectile. The engraving process plays a crucial role on determining the ballistic performance and projectile stability. This paper analyzes the dynamic response of a guided projectile during the engraving process. By considering the projectile center of gravity moving during the engraving process, a dynamics model is established with the coupling of interior ballistic equations. The results detail the stress situation of a guided projectile band during its engraving process. Meanwhile, the axial dynamic response of projectile in the several milliseconds following the engraving process is also researched. To further explore how the different performance of the engraving band can affect the dynamics of guided projectile, this paper focuses on these two aspects: (a) the effects caused by the different band geometry; and (b) the effects caused by different band materials. The time domain and frequency domain responses show that the dynamics of the projectile are quite sensitive to the engraving band width. A material with a small modulus of elasticity is more stable than one with a high modulus of elasticity.     

Effect of welding processes and consumables on fatigue crack growth behaviour of armour grade quenched and tempered steel joints

Quenched and Tempered (Q&T) steels are widely used in the construction of military vehicles due to its high strength to weight ratio and high hardness. These steels are prone to hydrogen induced cracking (HIC) in the heat affected zone (HAZ) after welding. The use of austenitic stainless steel (ASS) consumables to weld the above steel was the only available remedy because of higher solubility for hydrogen in austenitic phase. The use of stainless steel consumables for a non-stainless steel base metal is not economical. Hence, alternate consumables for welding Q&T steels and their vulnerability to HIC need to be explored. Recent studies proved that low hydrogen ferritic steel (LHF) consumables can be used to weld Q&T steels, which can give very low hydrogen levels in the weld deposits. The use of ASS and LHF consumables will lead to distinct microstructures in their respective welds. This microstructural heterogeneity will have a drastic influence in the fatigue crack growth resistance of armour grade Q&T steel welds. Hence, in this investigation an attempt has been made to study the influence of welding consumables and welding processes on fatigue crack growth behaviour of armour grade Q&T Steel joints. Shielded metal arc welding (SMAW) and Flux cored arc welding (FCAW) were used for fabrication of joints using ASS and LHF consumables. The joints fabricated by SMAW process using LHF consumable exhibited superior fatigue crack growth resistance than all other joints.     

我国警卫工作改进历程考察与思考

在考察我国警卫工作改进历程的基础上,对警卫工作改进进行了深入思考。提出我国警卫工作改进始终坚持“既要保证安全,又不脱离群众”这一主题;始终坚持党委和政府的领导,坚持制度化、规范化建设;体现在路线警卫工作、住地警卫工作和现场警卫工作三个方面。     

A novel gradient approach for optimal design and sensitivity analysis of EWMA control charts

Conventional control charts are often designed to optimize out‐of‐control average run length (ARL), while constraining in‐control ARL to a desired value. The widely employed grid search approach in statistical process control (SPC) is time‐consuming with unsatisfactory accuracy. Although the simulation‐based ARL gradient estimators proposed by Fu and Hu [Manag Sci 45 (1999), 395–413] can alleviate this issue, it still requires a large number of simulation runs to significantly reduce the variance of gradient estimators. This article proposes a novel ARL gradient estimation approach based on integral equation for efficient analysis and design of control charts. Although this article compares with the results of Fu and Hu [Manag Sci 45 (1999), 395–413] based on the exponentially weighted moving average (EWMA) control chart, the proposed approach has wide applicability as it can generally fit into any control chart with Markovian property under any distributions. It is shown that the proposed method is able to provide a fast, accurate, and easy‐to‐implement algorithm for the design and analysis of EWMA charts, as compared to the simulation‐based gradient estimation method. Moreover, the proposed gradient estimation method facilitates the computation of high‐order derivatives that are valuable in sensitivity analysis. The code is written in Matlab, which is available on request. © 2014 Wiley Periodicals, Inc. Naval Research Logistics 61: 223–237, 2014     

一种基于语义关系的情感词典自动构建方法

构建英文情感词典研究相对成熟,形成了丰富可靠的词典资源。而针对中文的研究时间短,中文情感分析词典资源较少。借鉴现有可靠的英文词典资源,提出了基于语义关系的情感词典自动构建算法,算法先从HowNet的概念中进行中文义原和词语抽取及语义分析,再利用HowNet概念中DEF中英文属性值,在英文情感词典SentWordNet中进行义原和词语情感值查询,最后根据词语和义原之间的语义关系进行词语的情感值计算。算法直接利用现有的英文情感词典,无需人工标注,生成的情感词典记录了词语的语义关系、情感极性值等多种信息,弥补了现有词典的不足。评测实验结果表明,根据算法实现的情感词典相比其他词典在准确率接近的情况下,召回率和F值最高,取得了较好的评测性能。     

舰艇编队防空过程建模及分析

防空作战是大型水面舰艇编队的主要作战样式之一。运用PEPA(性能评价过程代数)方法对舰艇编队防空作战全过程进行有效建模,形式化描述了预警探测、情报传递、命令下达、防空拦截等作战主要过程。建立的PEPA模型具有层次化结构,考虑了要素间相互协作,体现了编队防空作战分布、并发的特点。通过对建立的PEPA模型进行性能指标选择和稳定状态分析,得到了不同因素对编队能力发挥的影响情况,获得了防空作战的基本要素组成,从而提供了一种解决舰艇编队问题的新方法。