变精度粗糙集与粗糙集属性约简特征比较

通过具体问题,阐明了经典粗糙集模型分类质量、相对正域、决策类下近似具有非单调递减特征,变精度粗糙集模型在约简过程中分类质量和相对正域会出现跳跃现象,约简过程具有不稳定性.需要针对3者分别建立模型,使属性约简变得多样化.     

粗糙集属性约简在高校评教中的应用

粗糙集属性约简就是在保持信息系统的分类和决策能力不变的条件下,删除不相关或不重要的属性,使原有系统得到简化。利用一种基于粗糙集的属性约简算法,通过计算属性在可分辨矩阵中出现的频率来定义属性的重要性,可使约简结果保持近似精度。实验表明该算法可以简化评教指标体系,降低信息系统所需存储空间。     

基于熵权-TOPSIS模型的大风浪条件下引航员登船水域选址

为保障大风浪条件下引航员登船安全、周边船和被引船的安全以及港口的安全高效运营,选取合理的引航员登船水域至关重要。综合考虑熵权法和TOPSIS法的优点,基于熵权-TOPSIS模型对大风浪条件下引航员登船水域进行优选研究。以大风浪条件下广州港引航员登船水域选址为例,运用该模型对3个备选方案进行评价,求得方案2的相对贴近度最大。该方案所述的引航员登船水域为大风浪条件下最优登船水域,且与实际选址一致。经实例验证,该模型较为可靠,实用性强。     

Nanocontainer aqueous systems for removing polymeric materials from marble surfaces: A new and promising tool in cultural heritage conservation

The present study dealt with the characterisation of marble surfaces coated by three kinds of polymeric material usually employed in stones conservation and with the evaluation of the removal of these coatings from the surfaces by means of new nanocontainer aqueous systems constituted of micellar solutions (MS) and oil-in-water microemulsions (MC). The polymeric coatings were also subjected to accelerated artificial ageing in order to detect possible effects associated to natural photochemical ageing. The aim of the work was the assessment of the performance of such nanocontainer aqueous systems taking acetone (Ac) as a reference “traditional” solvent for the removal procedure. The characterisation was carried out by means of Scanning Electron Microscopy (SEM) and Environmental Scanning Electron Microscopy (ESEM), Fourier Transform Infra-Red (FTIR) and Nuclear Magnetic Resonance (NMR) spectroscopies and the results obtained showed that the performance of these innovative methods based on nanocontainer aqueous systems can be considered highly promising in the view of a “green approach” to the conservation of cultural heritage.     

基于知识简洁度的粗集聚类方法

提出了基于知识简洁度的粗集聚类方法。首先计算对象集合在每个属性下的划分 ;然后在对初始划分进行合并时 ,引进了不可分辨度的概念 ,考虑了其他属性的影响 ,同时避免了两个对象之间取值相近的度量问题 ;在形成最终聚类结果时 ,引进了知识简洁度的定义 ,使得最终的聚类结果科学合理     

Impact absorption of tennis shoe-surface combinations

The aim of this study was to investigate, for typical shoes and surfaces used in tennis, the relative role of the shoe and surface in providing cushioning during running. Five test surfaces ranging from concrete to artificial turf were selected, together with two shoe models. Impact absorbing ability was assessed mechanically using drop test procedures and biomechanically using peak magnitude and rate of loading of impact force and peak in-shoe pressure data at the lateral heel. Differences in biomechanical variables between shoe-surface combinations were identified using a two-way ANOVA (p < 0.05). Mechanical test results were found to rank the surfaces in the same order regardless of the shoe model, suggesting that the surface is influential in providing cushioning. However, for all mechanical and biomechanical (p < 0.05) variables representing impact absorbing ability, it was found that the difference between shoes was markedly greater than the differences between surfaces. The peak heel pressure data were found to rank the surfaces in the same order as the mechanical tests, while impact force data were not as sensitive to the changes in surface. Correlations between mechanical and biomechanical impact absorption highlighted the importance of testing the shoe-surface combination in mechanical tests, rather than the surface alone. In conclusion, mechanical testing of the shoe-surface combination was found to provide a strong predictor of the impact absorbing ability during running if pressure data were used. In addition, for typical shoe-surface combinations in tennis, the shoe was found to have more potential than the surface to influence impact loading during running. Finally, in-shoe pressure data were found to be more sensitive than force plate data to changes in material cushioning.     

加权粗糙集算法的PDA图书馆最优选书策略

针对标准贝叶斯网络模型在PDA图书馆选书分析的应用中表现出误差较大的问题,本文提出了一种基于加权粗糙集优化贝叶斯网络的PDA选书模型,首先采用粗糙集的方式,重构属性集合,用新的属性集合代替原来的属性集合,然后采用属性序描述方法,这样得到的约简后的条件属性将按照对决策属性影响逐渐减小的顺序依次排列,为了减小此缺陷的影响,采用加权的方式对基于粗糙集的贝叶斯模型进行改进。仿真试验结果表明,基于加权粗糙集优化贝叶斯网络的PDA选书系统功能完善,并且通过实际选书案例证明其选书精确性较高。     

基于粗糙集的青年科技人才评价指标体系构建

在对青年科技人才准确定义的基础上,结合实地调研获取的大量数据建立了原始青年科技人才二级评价指标体系,通过使用粗糙集理论中的属性约简原理及重要度原理对原始指标进行了筛选及赋权,构建出包含4个一级指标和15个二级指标的青年科技人才核心指标体系,为青年科技人才的选拔、培养及考核提供了参考依据。     

识别核心专利的粗糙集理论模型

[目的/意义]研究应用粗糙集理论分析专利信息的可行性及其方法,建立识别核心专利的粗糙集理论模型。[方法/过程]综述国内外研究核心专利识别的相关文献,确定包含8个指标的核心专利识别指标体系,再用粗糙集理论分析专利信息数据,确定各指标的权重系数,按照计算的核心专利综合指数(CICP)大小将某特定技术领域的专利信息序列分为核心专利、关键专利、重要专利和一般专利。[结果/结论]以通信领域的976件专利数据为分析样本,运用本文核心专利识别模型识别出11件核心专利,其中包含获得中国专利奖的专利,核心专利识别结果在一定程度上表明核心专利识别模型的可行性和科学性;本文核心专利识别模型具有一般性,可应用于分析研究专利价值及其重要性。     

基于统计粗集模型的航空发动机维修成本组合预测方法

维修成本是航空公司为实现降耗增效关注的重点之一。本文运用统计粗集理论属性重要性判定方法,建立了发动机维修成本组合预测模型,这种方法充分利用了粗集理论"让数据说话"的优势。实例证明,建立的组合模型较之单一预测模型在拟合和预测精度上都有明显提高,这种方法也为航空公司合理制定维修计划和管理维修资金奠定了基础。