数据库集群技术浅析

在介绍数据库集群定义的基础上,重点阐述了数据库集群的两大体系结构：基于数据库引擎的集群技术和基于数据库网关（中间件）的集群技术,并对数据库集群的核心技术：负载均衡和失效转移进行了分析;最后对数据库集群的两大体系结构进行了比较。     

空气钻井潜孔钻头本体与钻齿的过盈分析

潜孔钻头是空气钻井中一种重要的破岩工具,钻齿镶嵌过程中的过盈量直接影响钻齿与钻头本体的接触强度.文章以(φ)311.1mm潜孔钻头用(φ)19mm钻齿为研究对象,建立了钻头本体与钻齿过盈配合的力学模型,并进行计算机模拟仿真.通过对过盈量分别为0.01、0.02、0.04、0.06、0.07、0.08、0.09、0.1mm时钻齿与钻头本体的接触分析可知,过盈量小于等于0.07mm时,钻头本体均为弹性变形;当过盈量为0.08mm时,钻头本体接触区域局部开始发生塑性变形;当过盈量达到0.1mm时,钻头本体沿齿眼的应力值接近材料的强度极限.根据分析结果并考虑塑形强化的影响,推荐(φ)311.1mm潜孔钻头(φ)19mm钻齿的安装过盈量为0.07～0.08 mm.论文的研究工作为空气钻头钻齿的科学装配提供了理论依据.     

空气潜孔锤在地铁降水井施工中的应用

针对城市密集居住区内施工占地困难,环保要求高,夜间禁止施工等不利条件,在地铁降水井的施工中成功应用了空气潜孔锤,介绍了空气潜孔锤的施工工艺以及在工期、环保、造价等方面与传统反循环成孔进行了分析比较,从而为今后类似工程施工提供指导。     

相关性视觉对抗贝叶斯个性化排序推荐模型

传统推荐模型存在“数据稀疏”问题，且异构特征间深层语义未有效挖掘。提出相关性视觉对抗贝叶斯个性化排序（Correlation Visual Adversarial Bayesian Personalized Ranking, CVABPR）推荐模型。首先，基于原始MovieLens数据集，在IMDB爬取对应电影海报图像，构建全新多模态数据集：MovieLens-100k-WMI和MovieLens-1M-WMI。其次，基于SENet模型提取一组具有互补性的异构图像特征，准确描述电影海报图像。然后，改进聚类典型相关性分析模型，深入挖掘异构SENet特征间的聚类典型相关性。基于该相关性优化视觉贝叶斯个性化排序模型，以精准刻画待推荐项目。最后，在模型中主动加入扰动因子，通过对抗学习来增强视觉特征鲁棒性，使推荐模型更稳定，从而生成高质量个性化推荐结果。为验证CVABPR模型，在新的多模态数据集上完成全部实验。实验结果表明：CVABPR模型在这两个数据集上都有效。在MovieLens-100k-WMI数据集上，其推荐的均值平均精度（Mean Average Precision, MAP）指标较最强基线提升3.802%；在MovieLens-1M-WMI数据集上，其推荐的MAP指标较最强基线提升4.609%。CVABPR模型优于主流基线。消融分析实验表明：相比改进的聚类典型相关性分析模型，对抗学习在推荐中发挥更重要作用。此外，在数据稀疏度更高的MovieLens-1M-WMI数据集上，CVABPR模型能获得更大幅度性能提升，“数据稀疏”问题得到有效缓解，且异构特征间的深层语义也得以充分利用，CVABPR模型具备较高应用价值。     

Characterisation of Pomegranate Juices from Ten Cultivars Grown in Turkey

Pomegranate juices obtained from ten pomegranate cultivars of Turkey were analysed for their sugars, organic acids and the other quality parameters including pH, total soluble solids, and titratable acidity. Statistically significant differences were determined among the cultivars for parameters analysed. Citric acid was the predominant acid of all the cultivars studied. Cultivar “Zivzik,” one of the most popular cultivar in Turkey, showed best scores for total soluble solids (16.9 ± 0.06 °Brix), glucose (84.18 ± 0.42 g/l) and fructose (83.34 ± 0.81 g/l). Pomegranate juices were classified into three groups according to principal component analysis and cluster analysis.     

植物基肉制品营养组成分析与评价

为充分了解国内外市售植物基肉制品营养品质现状,筛选了国内外30种市售植物基肉制品,对其营养组成、营养质量指数(INQ)和营养当量值(NEU)进行了分析,通过相关性分析、主成分分析和聚类分析的方法,建立了植物基肉制品营养素组成综合评价模型,并进行了分级分类评价。结果表明,市售植物基肉制品以植物基鸡块、植物基牛排等半成品为主,营养评价指标主要包括能量(140.85～1 076.00 kJ/100 g)、蛋白质(5.20～25.00 g/100 g)、脂肪(1.00～22.00 g/100 g)、碳水化合物(1.40～24.58 g/100 g)和钠元素(313.00～1 080.00 mg/100 g),有的还特别标识含有膳食纤维、钙元素和铁元素等。聚类分析发现,高蛋白(≥12.00 g/100 g)产品有23种,占比76.67%,以植物基牛排、植物基鸡块为主;低脂肪(≤3.00 g/100 g)产品有植物基香肠和植物基猪肉共2种,占比6.67%;低碳水化合物(≤5.00 g/100 g)产品有11种,占比约36.67%,包括植物基鸡块、植物基猪肉和植物基牛排等。相关性分析显示,植物基肉制...     

Assessing the association between phenolic compounds and the antioxidant activity of Brazilian red wines using chemometrics

The objective of this study was to evaluate the association among chemical parameters, the commercial value, and the antioxidant activity of Brazilian red wines using chemometric techniques. Twenty-nine samples from five different varieties were assessed. Samples were separated into three groups using hierarchical cluster analysis: cluster 1 presented the highest antioxidant activity towards DPPH (68.51% of inhibition) and ORAC (30,918.64 μmol Trolox Equivalents/L), followed by cluster 3 (DPPH = 59.36% of inhibition; ORAC = 25,255.02 μmol Trolox Equivalents/L) and then cluster 2 (DPPH = 46.67% of inhibition; ORAC = 19,395.74 μmol Trolox Equivalents/L). Although the correlation between the commercial value and the antioxidant activity on DPPH and ORAC was not statistically significant (P = 0.13 and P = 0.06, respectively), cluster 1 grouped the samples with higher commercial values. Cluster analysis applied to the variables suggested that non-anthocyanin flavonoids were the main phenolic class exerting antioxidant activity on Brazilian red wines.     

花生主要农艺性状的相关性及聚类分析

应用DPS统计软件对18个花生品系的主要农艺性状的考种数据进行变异性、相关性和聚类分析。结果表明,主茎高、总分枝数、结果枝数、单株荚果数变异系数较大。主茎高与侧枝长、百果重与百仁重均呈极显著正相关;单株荚果产量与主茎高、侧枝长、总分枝数、结果枝数、单株果数、百果重、百仁重均呈正相关,与出米率呈负相关。18个花生品系因参与聚类的性状组合不同而聚类结果各异,综合性状的聚类在D＾2=5.91水平下分为4大类,其中第一类（1号和13号品系）和第四类（8、11、18）表现良好,值得推广种植。     

Classification of Iranian bottled waters as indicated by manufacturer’s labellings

Water is the most important substance in our daily lives and contains minerals which play an important role in our nutrition. In this study, the chemical composition of Iranian bottled water brands were investigated by correlation analysis, principal component analysis and hierarchical cluster analysis. For this purpose, the chemical composition reported on the label of 73 Iranian bottled waters was used as data set. It was found out that only 26 brands had eight important parameters such as calcium, magnesium, potassium, sodium, chloride, sulphate, bicarbonate and fluoride and 20 brands had acceptable charge balance error. Results showed that Iranian bottled waters can be divided into 11 classes. Most of them were Ca–Mg–HCO₃ type. The relationships among selected variables were also examined by Piper diagram. The best brands were introduced for common customers and kidney stone patients. The chemical content of Iranian bottled water brands was also compared with some world standards. It was observed that only one of the brands had fluoride in excess compare to that of standard values.     

防碰扫描方法在丛式钻井中的应用

采用丛式井开发低渗透油气田可以降低钻井成本和提高油气井产量,防碰绕障技术是丛式井钻井成败的关键。介绍了丛式井防碰扫描的原理方法、扫描参数的计算阱及具体应用,可以利用该方法对各井眼的空间关系及钻头运动趋势作出预测,从而避免因各种原因导致的井眼相碰,具有重要的现实意义。