一种高效的不协调决策表约简算法

目前,不协调决策表的分布约简、最大分布约简和分配约简算法复杂度较高,不适合处理大数据集。在分析已有算法基础上,分析了基于相对可区分度的属性重要性度量的性质,解决了正域度量属性重要性的缺陷。针对不协调决策表,给出了多种简化协调决策表的定义,从而大大缩减了约简的实例数。以相对可区分度为启发函数构造了一种高效完备的不协调决策表约简算法。理论分析和实验结果表明,该约简算法解决了现有算法在复杂度和属性重要性度量上的缺陷,适合处理不协调的大数据集。     

舰载光电系统粗跟踪分析

针对舰载激光武器光电跟踪系统对精度要求很高的问题,对光电跟踪系统粗跟踪误差源进行了详细的分析。论述了舰载激光武器光电跟踪系统对目标的跟踪过程,探讨了影响光电跟踪系统粗跟踪性能的各种因素,最后分析了舰船运动对跟踪误差的影响。文中给出了舰载光电跟踪系统粗跟踪的主要误差源,并介绍了一种稳定舰载光电跟踪仪视轴的有效方法。     

基于粗糙集理论的武器系统效能评定

简述了各种武器效能评定方法。建立武器参数效能模型,首先要挑选特征参数,这里采用知识约简方法选择武器的特征参数。利用神经网络理论建立了参数效能模型,武器系统的效能与武器特征参数的关系可通过神经网络的阈值和权值得到体现。通过实例对神经网络法与指数法所得的结果进行了比较,结果表明神经网络法比指数法精确。     

建筑材料燃烧性能分级新变革

介绍我国建筑材料燃烧性能分级的现状和未来的发展方向,提出引进欧盟的分级体系形成新的分级方法,并进行分析比较。     

一种基于VPRSM的辐射源模糊识别方法

为解决辐射源识别的问题,研究利用变精度粗糙集模型来确定辐射源识别特征权值的方法。由概念的正域中对象类的条件属性的描述,得到该概念的充分条件,并依据知识间的依赖性作为所要考虑的因素来获取特征权值,同时结合模糊集理论,提出一种基于变精度粗集特征加权的辐射源模糊识别方法。最后将此识别方法模型应用于雷达辐射源用途识别,仿真实验及其结果验证了该方法的可行性和有效性。     

Vapor-phase decomposition of dimethyl methylphosphonate (DMMP), a sarin surrogate,in presence of metal oxides

Chemical warfare agents (CWA) are stockpiled in large quantities across the globe. Agents stored in inaccessible facilities need to be destroyed rapidly without dispersing the compounds to surrounding areas. Metal-based energetic formulations are used in such prompt defeat applications to rapidly decompose the CWA by generating a high temperature environment. An alternate, and possibly a more effective decomposition pathway could be provided by chemicidal action of aerosolized condensed combustion products, which typically consist of metal oxides. Toxic fumes that escape the high tem-perature blast zone can be neutralized by smoke generated during combustion, depending on the par-ticle size, surface characteristics, chemical properties, and concentration of this smoke. This review considers relevant experimental and modeling studies quantifying decomposition of CWA comprising organophosphorus compounds and their surrogates on the surface of various metal oxides. Dimethyl methylphosphonate (DMMP), a sarine surrogate, was used most commonly for such experiments. Many reported efforts focused on the mechanisms of adsorption of DMMP to various metal oxides and initial reaction steps cleaving various bonds from the chemisorbed molecules. For selected oxides, these ex-periments were supported by quantum-mechanical calculations. In other studies, the capacity of oxide surfaces to adsorb and decompose DMMP was quantified. In most cases, porous catalysts were used although limited experimental data are available for aerosolized nonporous oxide particles. The reported experimental data applicable to scenarios involving prompt decomposition of CWA are summarized. It is noted that information is lacking describing respective heterogeneous reaction kinetics. Preliminary estimates of aerosolized smoke particle concentrations required to destroy CWA are made considering gas phase diffusion rates and reported values of the oxide capacity to decompose CWA or their surrogates.     

分解炉内煤粉燃烧和CaCO3分解流场的数值模拟

针对国内某大型水泥高性能分解炉,基于Fluent软件,采用有限速率/涡耗散模型模拟了炉内煤粉燃烧和生料分解的湍动多相流场,给出了炉内速度矢量、温度、压力和组分分布,结果显示,炉内流动趋势合理,可为分解炉的研究提供参考.     

基于粗糙集理论的加权排序法及其应用

针对粗糙集理论在有限个方案的多目标决策问题应用中存在的缺陷,提出了一种新的加权排序法,该方法既吸取了粗糙集约简属性的优点,又考虑到了各属性间的相对重要性,使得决策过程更符合实际情况.最后给出了该种排序法在武器装备系统论证中的应用,取得了较好的效果.     

基于FCM与粗糙集的连续数据知识挖掘方法

对连续数据知识挖掘可采用模糊C均值聚类将其离散化,再应用粗糙集进行约简.文中针对该过程中样本数据离散化后出现多个聚类中心隶属度值较接近的情况,提出了隶属度重叠度的方法来扩展对应离散类别的选取.离散类别确定后再应用粗糙集理论对其进行知识挖掘,可得到连续数据的本质特性.此方法应用于汽轮机轴系振动数据的知识挖掘.与采用最大隶属度确定对应类别的方法相比,该方法不影响对连续数据的知识挖掘,并能真实反映连续数据的特性.     

基于粗糙集和神经网络的联合模式识别系统

提出了一种划分属性离散区间的新方法.针对这种划分,提出一种约简和去噪的方法.随后,建立了粗糙集和LVQ神经网络的联合模式识别系统.最后,比较了用该系统和仅用神经网络进行识别的效果,证明了该方法的有效性.