关于内压缩流程空分设备液体产品“边际能耗”的讨论

引用经济学中"边际成本"的概念,提出了空分设备生产低温液体产品的"边际能耗"概念,计算出内压缩流程空分设备中液氧和液氮产品在各种情况下的边际能耗,并将计算结果与液化装置的液化单耗进行比较,提出适当增加内压缩流程空分设备的液体产量,可以提高空分设备的流程效率。     

基于PCA扩展的判别性特征融合

提出两个判别性的特征融合方法——主成分判别性分析和核主成分判别性分析。基于主成份分析和最大间隔准则理论,构造一个多目标规划模型作为特征融合的目标。随后,该模型被转化成一个单目标规划问题并通过特征分解的方法求解。此外,将一个近似分块对角核矩阵K分成c(c为数据集中的类别数)个小矩阵,并求出它们的特征值和特征向量。在此基础上,通过向量代数处理得到一个映射矩阵α,当核矩阵K投影到α上,同类样本的相似信息能最大程度地得到保持。本文中的实验证实两种方法的有效性。     

裂纹起始扩展阻力J_i与基本力学性能的关系

韧性裂纹开始扩展的J积分值对于韧性断裂的工程构件是一个重要的材料性能指标。统计了三个钢种两种焊缝 2 0 0多个J积分值。发现当裂纹生长量Δa <0 .6mm时 ,Δa与J积分值存在线性关系 ,可以用J =J0 +K(Δa)表示 ,式中K =di dΔa(Δa =a0 +da) ,a0 是J—Δa曲线与钝化线的交点。当da =0时的J值以Ji 表示 ,它表示了韧性裂纹开始扩展时的J积分值。实验证明 ,只有Ji 与静拉伸发生颈缩时的颈缩比功有良好的线性相关性 ,从而由Ji 换算得KⅠC也应与颈缩比功线性相关 (相关系数 >99% ) ,而与静力韧度无关。而韧性裂纹开始扩展以后 ,各个扩展量下的JR 值与静力韧度、颈缩比功无任何相关性。从J积分的条件、裂纹尖端的特征、断裂的微观机理对试验结果进行了分析。从工程应用角度出发 ,说明Ji 与屈服强度与强度极限之比 (称屈强比 )无关 ,低屈强比不是保证缺陷 (或微裂纹 )抵抗其扩展的有效方法 ,有效的方法是提高强度的同时 ,不降低均匀塑性变形能力Ψb,这可以通过提高冶金质量如细化晶粒来实现。     

The (In)efficiency of the retail-minus rule

The Retail-Minus Rule (RMR) specifies that the rate for a mandated wholesale service is set by subtracting from the vertically integrated provider's (VIP's) retail price the costs it avoids by selling the wholesale service rather than the retail service. While the RMR offers regulators an expedient method for setting wholesale prices, we show that it is efficient only under restrictive conditions. The proper test for a price squeeze inquires whether the VIP's retail price is greater than or equal to the sum of the direct cost and the opportunity cost of supplying the retail service. The RMR passes this test only when the competing retail services are perfect substitutes.     

软间隔组合凸线性感知器设计

组合凸线性感知器是用来构造分片线性分类器的一个通用理论框架。对于凸可分和叠可分情况,分别使用支持凸线性感知器算法和支持组合凸线性感知器算法将两类样本分开。在此基础上,文中提出一种软间隔的组合凸线性感知器设计方法。该方法首先映射原空间数据到高维特征空间,然后利用K均值算法将其中一类样本聚类成多个簇,并在每一簇与另一类样本间构造凸线性感知器,最后集成组合凸线性感知器。该方法能解决原感知器模型不适用非叠可分数据的问题,并且在一定程度上简化模型结构,在保证分类精度的前提下,提高泛化能力。实验结果证实文中方法的有效性,同其它分片线性分类器的对比也说明了它的优势。     

总间隔v-模糊支持向量机研究

为获得更好的分类性能,对传统模糊支持向量机(FSVM)进行扩展,提出一种总间隔v-模糊支持向量机(TM-v-FSVM)。通过使用差异成本及引入总间隔和模糊隶属度,同时解决不平衡训练样本问题和传统软间隔分类机的过拟合问题,从而提升学习机的泛化能力。采用UCI实际数据集进行模式分类实验,结果表明TM-v-FSVM具有稳定的分类性能。     

基于平均邻域最大化的手势识别方法研究

根据人机交互中手势控制系统的要求,提出一种基于平均邻域最大化(ANMM)算法的静态手势识别方法。将获得的二值化图像轮廓归一化到固定的解析度,构成多维向量,使用ANMM算法对同质相邻与异质相邻向量进行训练,计算出投影方向矩阵。将样本降维处理后,计算其在降维空间内与同质相邻和异质相邻向量的距离,判别样本所属分类。实验结果证明,该方法对静态手势的识别率可达90% 以上。     

基于加权最大类间边缘准则的人脸识别

在人脸识别中,特征提取技术被广泛应用于减少数据量和增强数据可分性。该文依据最大类间边缘准则,提出一种加权最大类间边缘准则的特征提取方法,引入加权函数,对类内和类间散布矩阵分别进行加权。并设计了一个基于离散小波分解、主成分分析和加权最大类间边缘准则的人脸识别系统。在ORL人脸库上的测试结果证实,该方法提高了识别率,最高识别率达98.25%。     

Determination and risk characterisation of bio-active piperine in black pepper and selected food containing black pepper consumed in Korea

Piperine is a bio-active compound found in pepper, including Piper nigrum L. and P. longum L. It has a strong, pungent flavour and several pharmacologic benefits. However, the risks of piperine have not yet been characterized. In this study, piperine in black pepper and some selected foods was determined to characterise the risk of exposure to piperine. Piperine in black pepper, curry and noodle was analysed by high-performance liquid chromatography–ultraviolet detection, which was validated through the measurement of performance parameters. The mean concentrations of piperine in black pepper, powdered curry, retorted curry, instant noodle and cup noodle were 4,418, 28, 3.4, 4.3 and 4.2 mg/100 g, respectively. The estimated dietary exposure to piperine was 123.66 μg/kg body weight/day, and the margin of exposure calculated by the no-observed-adverse-effect level of piperine, was 162. The piperine from food does not cause an adverse health effect to the public in Korea.     

Spatio-temporal metric learning for individual recognition from locomotion

Individual recognition from locomotion is a challenging task owing to large intra-class and small inter-class variations. In this article, we present a novel metric learning method for individual recognition from skeleton sequences. Firstly, we propose to model articulated body on Riemannian manifold to describe the essence of human motion, which can reflect biometric signatures of the enrolled individuals. Then two spatia-temporal metric learning approaches are proposed, namely Spatio-Temporal Large Margin Nearest Neighbor (ST-LMNN) and Spatio-Temporal Multi-Metric Learning (STMM), to learn discriminant bilinear metrics which can encode the spatio-temporal structure of human motion. Specifically, the ST-LMNN algorithm extends the bilinear model into classical Large Margin Nearest Neighbor method, which learns a low-dimensional local linear embedding in the spatial and temporal domain, respectively. To further capture the unique motion pattern for each individual, the proposed STMM algorithm learns a set of individual-specific spatio-temporal metrics, which make the projected features of the same person closer to its class mean than that of different classes by a large margin. Beyond that, we present a new publicly available dataset for locomotion recognition to evaluate the influence of both internal and external covariant factors. According to the experimental results from the three public datasets, we believe that the proposed approaches are both able to achieve competitive results in individual recognition.