基于Attention的AIS数据段相似性搜索模型

自动识别系统(Automatic Identification System, AIS)数据中蕴含着与船舶交通相关的时间序列信息。基于AIS数据段的相似性搜索也成为智能船运领域的研究热点。AIS数据属于多维时间序列数据,传统的多维数据序列相似性搜索方法通过将原始数据进行降维,并在降维后的数据空间进行时序数据的相似性搜索。然而,此种方法计算复杂度较高,会极大程度地降低相似性搜索的效率。深度学习模型产生的二进制编码改善了时间序列数据搜索性能。基于注意力(Attention)机制的AIS数据段相似性搜索模型对时间序列数据中的时序信息与参数变量间的潜在关系进行同时获取与学习。基于Attention机制的Seq2Seq的相似性搜索模型结合了时间序列数据与Seq2Seq模型的特点,使用Seq2Seq模型中GRU编码器产生的二进制编码进行AIS时间序列数据的相似性搜索。相应的实验结果也表明这种基于Attention机制的模型具有优异的AIS数据相似性搜索性能。     

灭火救援计算机模拟训练系统研究

从消防训练的角度入手,说明了建立灭火救援计算机模拟训练系统的必要性,通过与传统训练方式进行对比,介绍了灭火救援计算机模拟训练系统的特点,给出了基于GIS的灭火救援计算机模拟训练系统的框架、功能和特点;并在分析灭火救援计算机模拟系统缺陷的基础上,有针对性地提出了三级模拟训练体系,为消防训练提出了一个新的训练模式。     

建立完善红外光电产品技术标准 促进引导军民品装备及市场的发展

一．引言 随着红外热成像技术的发展及国外武器装备夜视装备的更新换代,我国海、陆、空、防化的武器装备均提出了装备二代热像仪的需求。红外热成像技术已经从一代发展到二代、三代,一代红外热成像装置采用红外探测器线阵扫描成像,二代和三代红外热成像装置则采用红外焦平面阵列,但总体上来说,二代和三代红外热成像装置仍处于发展阶段,相对于一代技术,二代热像仪的性能指标体系、     

The investigation of NTO/HMX-based plastic-bonded explosives and its safety performance

3-nitro-1,2,4-triazol-5-one (NTO) is the main component of insensitive munitions (IM) formulation because of its outstanding insensitive properties.In this paper,a series of NTO/HMX-based composite explosives were prepared and characterized.The study focuses on the effect of NTO on the performance of the formulations,especially the safety performance.The results revealed that the mechanical sensi-tivity of formulations was associated with NTO content,as well as the thermal conductivity,specific heat capacity and Arrhenius parameters.Then,the high amount of NTO using in formulation was proved to be helpful for NTO/HMX-based formulation to exhibit good thermal safety.Besides,by accelerating rate calorimeter (ARC) and a modified cook-offequipment,the pressure and pressure rise rate were proved as the important indicator for judging the thermal safety performance in confined spaces.Finally,the numerical simulation was used as a credible method for predicting the respond temperature of cook-off experiment.     

基于灰色区间关联决策的海上编队反导威胁评估

如何对舰艇编队的来袭导弹威胁程度进行科学的评估与排序,是防空反导指挥中的重要环节.目前威判中常常考虑最大值所提供的信息,存在结果失真的情况,所以采用区间数来表示更有现实意义.分析了舰艇编队多目标威胁评判的指标体系,据此给出了影响目标威胁的主要因素,并给出了主要因素的隶属函数,将区间数和灰色理论结合,构建了灰色区间关联决策模型并应用于多目标威胁排序中,实例计算分析表明,所提出的方法较好地解决了多目标多指标问题,对舰艇编队反导威胁判断提供了一个有效的参考处理方法.     

传热对水边界层流动的影响

本文评述了壁面传热对水边界层流体流动及其阻力的影响,同时分析了加热减阻的应用及其影响因素。传热不仅对层流流动、层流向湍流的转捩以及湍流流动有明显的影响,而且对流动分离以及由分离引起的空化也有显著影响。以热脉冲加热的主动式加热方式既可用来产生和消除边界层的T-S波,也能用来控制边界层的分离。因此,热控制技术可望在控制边界层转捩和分离,减小流动阻力等方面得到应用。     

Φ2mm丝材电弧喷涂工艺对涂层结合强度的影响

分析了喷砂预处理和电弧喷涂这两个过程的工艺参数对涂层的结合强度的影响。将分数维理论用于喷砂后的试样表面,研究了喷砂角度、分数维和结合强度的相互关系,找到了获得最高结合强度所用的喷砂角度;通过正交试验法,找到了能获得最佳涂层结合强度的Φ2mm丝材电弧喷涂工艺参数。     

频率捷变雷达干扰方法分析

频率捷变雷达有许多其它雷达无法比拟的优点,在现代雷达体制中占有重要地位。分析了频率捷变雷达的特点,根据其工作原理,提出了干扰频率捷变雷达的几种方法,分析了各自的优点和局限性。阐述了干扰方法的原理及实现的技术途径,并给出了有关参数的计算方法和实例。     

Minimizing total completion time in two-processor task systems with prespecified processor allocations

We consider the problem of scheduling multiprocessor tasks with prespecified processor allocations to minimize the total completion time. The complexity of both preemptive and nonpreemptive cases of the two-processor problem are studied. We show that the preemptive case is solvable in O(n log n) time. In the nonpreemptive case, we prove that the problem is NP-hard in the strong sense, which answers an open question mentioned in Hoogeveen, van de Velde, and Veltman (1994). An efficient heuristic is also developed for this case. The relative error of this heuristic is at most 100%. © 1998 John Wiley & Sons, Inc. Naval Research Logistics 45: 231–242, 1998     

Transactional and in-store display data of a large supermarket for data-driven decision-making

This article describes daily and monthly transactional and in-store display data of a large supermarket from January to October in 2019 associated with 28 757 stock-keeping units (SKUs) in 5 categories and 41 subcategories. The database contains five parts, including information about each SKU, in-store display, daily sales, inventory, and replenishment. We also propose some research questions related to assortment planning, pricing, inventory management, and customer behavior. Researchers are welcome to develop data-driven models or other innovative methods to address these questions or other practical problems using this database.