基于相对知识粒度的区间集决策信息表不确定性度量

区间集决策信息表拓展了经典决策信息表,其粒化结构的不确定性刻画成为重要应用基础。基于区间邻域粒化引入相对知识粒度,进而提出区间集决策信息表的一种新型不确定性度量。利用两对象之间的距离及半径来建立区间邻域粒化系统,证明距离粒化与相似度刻画的条件等价性,获得区间近似粗糙度;把经典相对知识粒度推广为区间相对知识粒度,将其与区间近似粗糙度进行信息融合,提出一种新型不确定性度量并得到了粒化单调性等性质。最后进行了实例验证,结果表明所提度量能够有效表征粒化结构变化所引起的不确定性变化。

肿节风颗粒特定条件下最优制粒区间研究

目的:初步探索获得中药制剂最优制粒区间的可行性。方法:以肿节风颗粒成型工艺为模型,采用析因设计进行试验,以颗粒得率、颗粒成型率加权后的综合评分值为考察指标,采用Origin 8.0软件分别绘制不同含水量混合粉(浸膏粉-可溶性淀粉)综合评分值与润湿剂浓度、液固比关系效果图,并用Origin 8.0软件中的Screen Reader功能进行坐标提取,综合得到肿节风颗粒在特定条件下的最优制粒区间;在该制粒区间下,选择不同批次浸膏粉进行实验室规模及处方放大试验进行验证。结果 :当浸膏粉-可溶性淀粉配比为1∶2时,肿节风颗粒存在一个最适制粒区间,即当物料含水量在[2.0%,6.0%]、润湿剂浓度在[50%,80%]及液固比在[15.3 mL/g,18.6 mL/g]范围内时,所制肿节风颗粒均符合要求;验证试验的综合评分值均大于85%,证实所制颗粒合格。结论:本研究建立的考察中药浸膏在特定条件下的最优制粒区间的方法,对解决中药制粒工业生产批间差异大、重现性差等问题,具有一定的参考作用。

区间直觉模糊粒结构的距离度量

在粒结构不确定性的度量上区间直觉模糊信息粒度存在局限性.基于此,提出区间直觉模糊粒结构距离的概念.首先,给出区间直觉模糊粒结构距离的定义,其次,通过定理的证明和实例的分析来验证该距离的有效性,最后,就其潜在的应用进行了探讨.

基于粒计算的离散化算法及其应用

连续数值属性的离散化是粒计算理论应用的重要步骤。首先对目前的离散化算法进行了分类讨论,提出了区间粒的概念,融合熵理论定义了区间粒的粒度,进而提出了基于粒计算的连续数值属性的离散化算法,并将该算法应用于入侵检测过程;实验结果表明该算法简洁高效,能够确保入侵检测系统的检测效果。

利用源域结构的粒迁移学习及词性标注应用

迁移学习在一定程度上减轻了目标域的数据稀疏问题对泛化能力的影响,然而泛化能力的提高仍然受到负迁移等问题的影响。为了解决负迁移问题,该文提出使用源域结构的文本语料的信息粒化方法,用区间信息粒表示出源域数据集的结构对数据集中统计量的影响。然后提出区间二型模糊隐马尔可夫模型(Interval Type-2fuzzy Hidden Markov Model,IHMM)以处理区间信息粒。给出了IHMM的构建方法和去模糊化方法。在文本的词性标注任务中进行了多个实验,可以证实利用源域结构信息的粒迁移学习方法避免了负迁移,提高了模型的泛化能力。

一种新颖的离散化算法及其应用

连续数值属性的离散化是粒计算理论应用的重要步骤。首先对目前的离散化算法进行分类讨论,提出区间粒的概念,融合熵理论定义区间粒的粒度,进而提出基于粒计算的连续数值属性的离散化算法,并将该算法应用于入侵检测过程。实验结果表明该算法简洁高效,能够确保入侵检测系统的检测效果。

区间值犹豫模糊信息系统中的信息粒和粗糙熵

信息粒和粗糙熵作为研究信息系统不确定性问题的两种主要方法,已被广泛应用于许多领域。本文基于区间值犹豫模糊二元关系,给出了区间值犹豫模糊粒度结构概念和区间值犹豫模糊粒的基数概念,讨论了区间值犹豫模糊粒度结构上的三种偏序关系。在区间值犹豫模糊粒度结构基础上给出了区间值犹豫模糊信息粒度和粗糙熵的概念,讨论了区间值犹豫模糊信息粒度和粗糙熵的偏序关系,分析了区间值犹豫模糊信息粒度和粗糙熵与已有信息粒度和粗糙熵之间的关系,并通过实例验证了有关定义和定理的正确性。

Optimizing Structure and Processes of Nickel Induced Lateral Crystallization

The structure and processes of nickel induced lateral crystallization are studied.The structure of metal induced lateral crystallization(MILC) is improved by opening a seed window on the buried oxide,which is helpfu to get superior quality of large grain poly Si at low temperature.By optimizing the temperature and time of annealing based on others' pervious work,the large grain poly Si with few defects are obtained,and the typical grain size is 70～80μm.The methods of etching NiSi 2 which is created after the long time annealing are also studied for the first time.Finally,a method is successfully chosen to reduce the possible contamination of Ni and to guarantee the MILC for the submicron VLSI application.

Effect of Ventilation Strategies on Particle Distribution in a Two-Zone Ventilated Room

A computational fluid dynamic （ CFD ） analysis of air movement and aerosol particle transport in a two-zone ventilated room with an inter-zonal opening is presented to study the impact of ventilation strategies and size of the opening on indoor particle dispersion and concentration distribution. The comparisons of average particle concentrations in both zones between the computations and the experiments from the literature are generally satisfactory and acceptable. The combined effects of sizes of the opening and the inlet and outlet locations （three different strategies） are simulated and discussed. The results show that ventilation strategy and size of the opening influence the particle removal rate in zone 1. The removal rate is decreased when the air supply system is changed from the tap-inlet to the bottom-inlet configuration. The top-inlet system obtains a better particle deposition in zone I than the bottom-inlet configuration. However, the particle concentration at breathing level is lower for bottomsupply system than for top-supply. Decreasing the size of interzonal opening increases the particle deposition rate in zone 1 only for the top.supply system, especially for coarse particles.

A common origin of muon g-2 anomaly,Galaxy Center GeV excess and AMS-02 anti-proton excess in the NMSSM

The supersymmetric model is one of the most attractive extensions of the Standard Model of particle physics.In light of the most recently reported anomaly of the muon g-2 measurement by the FermiLab E989 experiment,and the excesses of gamma rays at the Galactic center observed by Fermi-LAT space telescope,as well as the antiproton excess observed by the Alpha Magnetic Spectrometer,we propose to account for all these anomalies or excesses in the Next-to-Minimal Supersymmetric Standard Model(NMSSM).Considering various experimental constraints including the Higgs mass,B-physics,collider data,dark matter relic density and direct detections,we find that a~60 GeV bino-like neutralino is able to successfully explain all these observations.Our scenario can be sensitively probed by future direct detection experiments.

An adaptive smoothed particle hydrodynamics for metal cutting simulation

Normally large amounts of particles are required to accurately simulate the metal cutting process,which consumes a lot of computing time and storage.Adaptive techniques can help decrease the number of particles,hence reducing the runtime.This paper presents a novel adaptive smoothed particle hydrodynamics(SPH)method for the metal cutting simulation.The spatial resolution changes adaptively according to the distance to the tool tip by the particle splitting and merging.More particles are selected in the region where the workpiece and the tool are in contact.Since the contact region constantly changes during the cutting process,two quadrilateral frames are adopted in the adaptive algorithm to dynamically change the distribution of particles.One frame for the refinement,the other for the coarsening.These frames move at the same speed as the tool.To test the computational efficiency,the metal cutting process is simulated by using SPH with three different adaptive approaches.Numerical results show that the proposed adaptive algorithm with dynamic refinement and coarsening can significantly optimize the runtime.