三维荧光二阶校正法直接快速测定血浆样中川芎嗪和阿魏酸含量

本文将三维荧光光谱与二阶校正方法相结合用于人体血浆样中的中药药理活性成分川芎嗪和阿魏酸含量的直接测定.尽管预测样存在血浆内源荧光物质的基体干扰,但由于该方法基于“数学分离”的思路和具有“二阶优势”,仍能对荧光光谱严重重叠的血浆样中的目标分析物进行直接同时定量测定,为川芎嗪和阿魏酸在体内的代谢监控及其药效研究提供了一种新思路.获得的预测样中川芎嗪和阿魏酸的平均回收率分别为(95.7±2.2)％和(100.8±2.5)％.该方法简单、快速,结果可靠.     

二阶校正法与三维荧光光谱相结合直接测定人体血浆中的水杨酸

水杨酸、萘普生及血浆背景干扰组分,这三者荧光光谱重叠严重,如果不经分离,无法用常规方法定量分析水杨酸.本文利用二阶校正的优势,将三维激发发射荧光光谱与平行因子分析算法和交替三线性分解算法相结合,快速分析测定血浆中的水杨酸.当因子数选为4时,用两种算法获得的回收率分别为(103.8±3.5)%和(98.4±1.4)%.实验结果表明,此方法可用于干扰组分共存下,快速定量测定血浆中的水杨酸,从而可用于研究阿斯匹林的代谢过程.     

激发发射荧光矩阵二阶校正方法直接测定人体血浆样中多沙唑嗪含量

多沙唑嗪是一种α-1受体阻滞剂,有选择性地作用于α-1受体,可改善良性前列腺增生(BPH)症状,又可降低血压.化学计量学二阶校正算法具有独特的"二阶优势",能够在未知干扰组分共存下,直接定量分析体系中感兴趣的组分.本文基于多沙唑嗪稳定的荧光性质,利用激发发射荧光矩阵(EEMs)结合二阶校正算法,成功地直接测定血浆样中多沙唑嗪.血浆样品不需做前期处理.它快速简便,比其它分析方法,既节约试剂又节约时间,是一种对复杂体系感兴趣组分进行直接定量分析的好工具.     

三维荧光光谱结合二阶校正算法用于化妆品中芦丁含量的测定

本文利用2种分别基于交替三线性分解算法(ATLD)和自加权交替三线性分解算法(SWATLD)的二阶校正算法与三维荧光光谱相结合,对化妆品中的芦丁进行直接定量分析.当选取组分数为3时,ATLD和SWATLD获得的平均回收率分别为(94.4±2.7)%和(100.2±1.1)%.另外,采用品质因子,如灵敏度(SEN)、选择性(SEL)和检测下限(LOD)评估了这2种算法所得结果的准确性.实验表明这2种算法能成功地用于分析化妆品中芦丁的含量,而且在这个体系中,SWATLD的性能较ATLD稍优.     

三维荧光结合交替三线性方法测定中药升麻和人体血浆中升麻素的含量

本文利用化学计量学交替三线性分解(ATLD)方法解析三维荧光数据,在有未知干扰组分共存的情况下,对中药升麻和人体血浆中升麻素的含量和回收率进行了测定,利用核一致诊断法判断得到的组分数是2。中药升麻和血浆体系的背景与升麻素有严重的重叠,用ATLD方法得到的解析光谱与真实光谱几乎完全重合,定量解析结果也令人满意。测定结果显示,标准曲线法和二阶标准加入法得到的中药升麻中升麻素的含量分别为0.061±0.003%和0.055±0.002%,回收率为89.8±4.7%,人体血浆中升麻素的回收率分别为102.7±1.6%。实验结果表明,此法可用于干扰组分共存下中药升麻和人体血浆中升麻素的快速准确定量分析。     

近红外光谱法测定黄芩提取物中黄芩苷含量

应用近红外光谱技术探讨了黄芩提取物中黄芩苷含量的快速检测新方法.共收集12个不同厂家的100批黄芩提取物样品,利用Nicolet 6700型傅立叶变换近红外光谱仪采集样品近红外漫反射光谱,通过HPLC法测定黄芩苷含量值,结合偏最小二乘法(PLS)建立黄芩苷含量的近红外定量校正模型,并通过筛选合适的光谱预处理方法、建模区间和主成分数获得最优模型.所建最优校正模型的相关系数(R2)、校正均方差(RMSEC)和内部交叉验证均方差(RMSECV)分别为0.995、0.440和2.259;经外部验证,模型的预测相关系数(r2)、预测均方差(RMSEP)和平均回收率分别为0.988、0.486和100.190%.结果表明,该方法可用于不同厂家黄芩提取物中黄芩苷含量的直接测定,操作简便,无污染,结果准确可靠,可实现大批量样品的快速分析.     

NIRS快速测定制何首乌中二苯乙烯苷的含量

目的建立近红外光谱技术(NIRS)测定制何首乌中2,3,5,4'-四羟基二苯乙烯-2-O-β-D-葡萄糖苷(二苯乙烯苷)含量的方法。方法中国药典方法测定样品中二苯乙烯苷的含量,采集样品的近红外光谱数据,利用TQ 8.0分析软件对光谱数据进行分析处理,建立制何首乌中二苯乙烯苷含量测定的近红外光谱校正模型。结果该模型的内部交叉验证系数(R~2)、校正均方差(RMSEC)、预测均方差(RMSEP)、内部交叉验证均方差(RMSECV)分别为0.98、0.04、0.04、0.12。结论所建立的方法可用于制何首乌中二苯乙烯苷的含量测定。     

NIRS结合PLS快速分析银黄颗粒中有效成分含量

研究旨在探讨利用银黄颗粒样品的近红外漫反射光谱(NIRS)信息,建立黄芩苷和绿原酸含量的校正模型,为银黄颗粒质量的快速评价提供1种新方法.以HPLC分析值为参照,采用近红外漫反射光谱技术采集100批银黄颗粒样品的近红外漫反射光谱,结合偏最小二乘法(PLS)建立了黄芩苷和绿原酸含量的校正模型.黄芩苷和绿原酸含量的校正模型相关系数(R2)分别为0.998和0.995,校正均方差(RMSEC)为0.578和0.123,内部交叉验证均方差(RMSECV)为2.356和0.412;经外部验证,预测相关系数(r)分别为0.995和0.984,预测均方差为(RMSEP)0.597和0.166.结果表明,该方法准确、简便、无污染,可实现大批量银黄颗粒样品的快速分析.     

高良姜中高良姜素含量NIRS分析模型的构建

建立高良姜药材中高良姜索含量的近红外光谱定量分析模型,实现药材中高良姜素的含量的快速测定方法,采用高效液相色谱法测定130批药材中高良姜素的含量,采集近红外光谱数据并用多元散射校正法、二阶导数法、Savitzky-Golay平滑法预处理,结合偏最小二乘法建立高良姜中高良姜素含量的定量分析模型,对所建模型进行了内部交叉验证和外部预测验证,并对模型进行了重复性考察。对于所建立的高良姜素近红外光谱定量分析模型,内部交叉验证决定系数达到0.9868,校正均方差为0.0529,预测均方差为0.0625,内部交叉验证均方差为0.0975,交叉检验和外部检验RPD均大于3。表明该模型稳定、准确可靠,可应用于高良姜中高良姜素的含量测定。     

二阶变分图像恢复模型的重启动快速ADMM方法

目的 基于二阶导数的图像恢复变分模型可以同时保持图像边缘与光滑特征,但其规则项的非线性、非光滑性,甚至非凸性制约着其快速算法的设计。针对总拉普拉斯(total Laplacian,TL)与欧拉弹性能(Euler’s elastica,EE)两种图像恢复变分模型,在设计快速交替方向乘子法(fast alternating direction methods of multipliers,fast ADMM)的基础上引入重启动策略,以有效消除解的振荡,从而大幅提高该类模型计算效率,并为其他相近模型的快速算法设计提供借鉴。方法 基于原始ADMM方法设计反映能量泛函变化的残差公式,在设计的快速ADMM方法中根据残差的变化重新设置快速算法的相关参数,以避免计算过程中的能量振荡,达到算法加速目的。结果 通过大量实验发现,采用原始ADMM、快速ADMM和重启动快速ADMM算法恢复图像的峰值信噪比(peak signal-to-noise ratio,PSNR)基本一致,但计算效率有不同程度的提高。与原始ADMM算法相比,在消除高斯白噪声和椒盐噪声中,对TL模型,其快速ADMM算法分别提高6%50%和13%240%;重启动快速ADMM算法提高100%433%和100%466%;对EE模型,其快速ADMM算法分别提高14%54%和10%83%;重启动快速ADMM算法分别提高100%900%和66%800%。此外,对于不同的惩罚参数组合,相同模型的快速ADMM算法的计算效率基本相同。结论 对于两种典型的二阶变分图像恢复模型,本文提出的快速重启动ADMM算法比原始ADMM算法及快速ADMM算法在计算效率方面有较大提高,计算效率对不同惩罚参数组合具有鲁棒性。本文设计的算法对于含其他形式二阶导数规则项的变分图像分析模型的重启动快速算法的设计可提供有益借鉴。     

基于平行因子分解的协同聚类推荐算法

针对三元组数据内在关联性复杂的特点,提出了基于平行因子分解(PARAFAC)的协同聚类推荐算法。该算法利用PARAFAC算法对张量进行分解,挖掘多维数据实体之间的相关联系和潜在主题。首先,利用PARAFAC分解算法对三元组张量数据进行聚类;然后,基于协同聚类算法提出了三种不同方案的推荐模型,并通过实验对三种方案进行了比较,得到了最优的推荐模型;最后,将提出的协同聚类模型与基于高阶奇异值分解(HOSVD)的推荐模型进行比较。在last.fm数据集上,PARAFAC协同聚类算法比HOSVD张量分解算法在召回率和精确度上平均提高了9.8个百分点和3.7个百分点,在delicious数据集上平均提高了11.6个百分点和3.9个百分点。实验结果表明所提算法能更有效地挖掘出张量中的潜在信息和内在联系,实现高准确率和高召回率的推荐。     

Tensor-based blind alignment of MIMO CD processes

A key issue in paper–machine cross-directional (CD) control is alignment, i.e., accurate spatial mapping of the actuators to the scanning points. Typically, this mapping problem is a non-linear and slowly time-varying phenomenon. Most current methods require bump tests, in which a few actuators are excited, and the peaks in the observed scan data are assigned to the excited actuators. A major drawback of these methods is that they need to be manually initiated and thus require the CD control system to be in manual mode. This paper presents a novel, deterministic, tensor-based modeling of the CD process and an alignment method that works while the closed-loop CD controlled system is running. First, we link the CD data to the parallel factor (PARAFAC) model. Exploiting this link, we derive a deterministic blind PARAFAC decomposition as an alignment method with performance close to non-blind minimum mean-square error (MMSE). We show that blind alignment follows from simultaneous matrix decomposition. The proposed PARAFAC capitalizes on the physical location of the actuators, scanning databoxes and their temporal diversities. Its performance is verified in several simulations for different actuator models. The discussed algorithm is then tested on industrial paper–machine data and evaluated as an identification tool.     

Joint frequency, 2-D DOA,and polarization estimation using parallel factor analysis

This paper proposes a new algorithm for joint frequency, two-dimensional (2-D) directions-of-arrival (DOA), and polarization estimation using parallel factor (PARAFAC) analysis model and cumulant. The proposed algorithm designs a new array configuration, and extends the PARAFAC analysis model from the common data-domain and subspace-domain to the cumulant one, and forms three-way arrays by using the three cumulant matrices obtained from the properly chosen dipole outputs, and analyzes the uniqueness of low-...     

Algorithms for sparse nonnegative Tucker decompositions

There is a increasing interest in analysis of large-scale multiway data. The concept of multiway data refers to arrays of data with more than two dimensions, that is, taking the form of tensors. To analyze such data, decomposition techniques are widely used. The two most common decompositions for tensors are the Tucker model and the more restricted PARAFAC model. Both models can be viewed as generalizations of the regular factor analysis to data of more than two modalities. Nonnegative matrix factorization (NMF), in conjunction with sparse coding, has recently been given much attention due to its part-based and easy interpretable representation. While NMF has been extended to the PARAFAC model, no such attempt has been done to extend NMF to the Tucker model. However, if the tensor data analyzed are nonnegative, it may well be relevant to consider purely additive (i.e., nonnegative) Tucker decompositions). To reduce ambiguities of this type of decomposition, we develop updates that can impose sparseness in any combination of modalities, hence, proposed algorithms for sparse nonnegative Tucker decompositions (SN-TUCKER). We demonstrate how the proposed algorithms are superior to existing algorithms for Tucker decompositions when the data and interactions can be considered nonnegative. We further illustrate how sparse coding can help identify what model (PARAFAC or Tucker) is more appropriate for the data as well as to select the number of components by turning off excess components. The algorithms for SN-TUCKER can be downloaded from M?rup (2007).     

基于可视化传感器阵列系统的肺癌标志物快速检测研究

肺癌的早期快速诊断对于肺癌患者的治疗至关重要。针对肺癌患者所呼出的特定标志物,建立可视化传感器阵列系统,对4种肺癌标志物进行了实验研究。采用分层聚类分析、主成分分析的统计学方法对检测结果进行分析。对不同肺癌标志物、不同体积分数的样本在聚类分析中可以正确分类,且结构相似体积分数相近的样本能优先聚到一簇。利用主成分分析获得的前2个主成分所代表的肺癌标志物72.0%的信息量即可以实现不同类标志物样本区分。研究表明:这种可视化传感器阵列系统是一种快速有效的检测识别肺癌标志物的方法。     

多层极限学习机在入侵检测中的应用

针对神经网络在入侵检测应用存在的维度高、数据大、获取标记样本难、特征构造难、训练难等问题,提出了一种基于深度多层极限学习机(ML-ELM)的入侵检测方法。首先,采用多层网络结构和深度学习方法抽取检测样本最高层次的抽象特征,用奇异值对入侵检测数据进行特征表达;然后,利用极限学习机(ELM)建立入侵检测数据的分类模型;其次,利用逐层的无监督学习方法解决入侵检测获取标记样本难的问题;最后采用KDD99数据集对该方法的性能进行了验证。实验结果表明:多层极限学习机的方法提高了检测正确率,检测漏报率也低至0.48%,检测速度比其他深度模型的检测方法提高了6倍以上。同时在极少标记样本的情况下仍有85%以上的正确率。通过多层网络结构的构建提高了对U2L、R2L这两类攻击的检测率。该方法集成深度学习和无监督学习的优点,能对高维度,大数据的网络记录用较少的参数得到更好的表达,在入侵检测的检测速度以及特征表达两个方面都具有优势。     

一种基于LS-SVM的特征提取新方法及其在智能质量控制中的应用

提出一种基于最小二乘支持向量机(LS-SVM)的特征提取新方法,并将其成功应用到智能质量控制领域。首先,将线性特征提取公式表达成与LS-SVM回归算法中相同的形式;再遵循SVM方法将数据集由原输入空间映射到高维特征空间,进而使用该技巧通过线性形式实现非线性特征提取。然后,用常规控制图提取出一个含有6种模式、50维特征的仿真数据集用于测试,通过LS-SVM特征提取后,原数据集的特征被降到了3维并保留了原80%的分类信息。最后,用BP分类器对特征提取后的样本进行识别,其结果优于新型RSFM网络直接对原始样本进行识别的效果。仿真实验结果表明了LS-SVM特征提取方法的可行性和有效性。     

基于关系网络的轴承剩余使用寿命预测方法

针对轴承全寿命周期数据获取困难、训练样本少的问题, 提出一种基于关系网络的轴承剩余使用寿命(Remaining useful life, RUL)预测方法. 关系网络是一种基于度量的元学习方法, 在少量训练样本下, 具有快速学习新任务的优点. 设计了一种基于关系网络的轴承健康评估模型, 利用关系网络的嵌入模块提取轴承状态特征, 利用关系模块度量轴承状态特征之间的相似性, 基于相似性构建轴承健康指标(Health indicator, HI); 对健康指标进行Savitzky-Golay滤波平滑处理, 降低振荡对预测结果的影响; 最后利用线性函数对健康指标进行拟合, 得到轴承RUL预测值. 为验证所提方法的有效性, 在PHM2012轴承实测数据集上进行实验. 结果表明, 所得健康指标能够反映轴承的退化趋势, 所得RUL预测结果与空间卷积长短期记忆神经网络 (Convolutional long short-term memory neural network, ConvLSTM)、Transformer、循环神经网络(Recurrent neural network, RNN)、卷积神经网络(Convolutional neural network, CNN) + 长短期记忆网络 (long short-term memory network, LSTM )、编码器−解码器(Encoder-decoder) + 注意力机制 (Attention mechanism)方法相比, 误差百分比分别减少了1.67%, 3.40%, 9.02%, 13.71%, 30.48%. 该方法在少量训练样本的基础上可以取得较好的预测结果, 具有一定的应用价值.     

Temporal and Spatial Variation of Vegetation Phenology in Temperate China and Its Impact on Gross Primary Productivity

Land surface phenology is defined as the seasonal timing of life cycle events of vegetated land surface on local or global scale.Most studies of vegetation phenology in China’s temperate zone are focused on single vegetation type in certain area，the studies about long-time vegetation phenology on large scale is rare.The influence of vegetation phenology on GPP(gross primary productivity) remains to be determined.Using Moderate Resolution Imaging Spectroradiometer(MODIS) MCD12Q2 data from 2001 to 2014，start of growing season(SOS)，end of growing season(EOS) and length of growing season(LOS) in temperate China(>30°N) are obtained.GPP from MODIS MOD17A3 data for the same period is also obtained.Using regression analysis and correlation analysis methods，spatial and temporal patterns of SOS，EOS and LOS are analyzed.The impacts of SOS，EOS and LOS on interannual variability of GPP are also analyzed.Results show that the average and standard deviation of SOS，EOS and LOS from 2001 to 2014 are 121±10，270±12 and 153±12 days，respectively.The trend of earlier SOS，delayed EOS and increased LOS are not significant(p>0.05)，but LOS shows positively correlated to GPP.The spatial distribution of annual average LOS and GPP from 2001 to 2014 presents an increase trend from northwest to southeast.Regions with significant interannual variation(p<0.05) of SOS，EOS and LOS are 13%，21% and 13.2%，respectively.Regions of significant correlation(p<0.05) of SOS，EOS and LOS to GPP account for 8.31%，9.33% and 8.72% of the study area.GPP has mainly medium correlations(p<0.05，0.5<|r|<0.8) to SOS，EOS and LOS.     

The development of multiple reaction monitoring assays for liver-derived plasma proteins

There is wide interpatient variability in toxicity to chemotherapeutic drugs and a lack of routine clinical tests for prospectively identifying patients at risk of developing toxicity from chemotherapy. An empirically driven MS strategy has been developed to monitor liver-derived plasma proteins as potential biomarkers of early toxicity. Multiple reaction monitoring (MRM) has been used to assess 46 candidate peptides from 18 liver-derived proteins. Following an iterative process of assay design, optimisation and assessment we selected 29 MRM assays (median CV 4.6%, range 1.2-11.6%) and monitored changes in levels of plasma proteins from a small number of colorectal cancer (CRC) patients undergoing chemotherapy. We demonstrated MRM assay robustness, and show that patients undergo minor elevation in plasma proteins when profiled on Day 3 of the chemotherapeutic regime. The MRM assays were in general agreement with 2-D DIGE-based quantitation from the same patient samples. The data supports the application of MRM-based methods as facile, highly reproducible, medium-throughput techniques that warrant expanded investigation for clinical utility in identifying patients at risk of developing chemotoxicity.