关节软骨的红外光谱成像及支持向量机定量研究

傅里叶变换红外光谱成像(FTIRSI)技术可同时获得样本组织的显微结构信息和红外光谱信息,结合化学计量学算法可用于样品光谱信息的定量分析和组织判别的研究。基于此,本实验采用FTIRSI技术结合支持向量机分类(SVC),实现健康和病变关节软骨组织的判别,其识别率分别为100.0%和95.4%,准确率达到97.7%;基于FTIRSI建立支持向量回归(SVR)模型定量研究关节软骨样本中胶原蛋白和蛋白多糖两种生物大分子的含量与分布,发现病变关节软骨中蛋白多糖的含量发生流失,尤其在表层区。研究结果表明,FTIRSI与支持向量机(SVM)相结合有望成为一种新型的骨关节炎诊断工具,对骨关节炎监测和诊断研究具有重要意义。     

基于核函数的非线性分类相关分析及其在化学模式识别中的应用

与统计分析和神经网络相比,基于结构风险最小的支持向量机有更好的分类性能。它用于非线性分类时,先将样本映射到更高维的特征空间,往往会增加复共线性与冗余信息,将影响样本分布,降低线性支持向量机分类器(LSVC)的预测性能。本研究提出非线性分类相关分析算法(NLCCA),利用核函数技术,无需了解非线性映射的算式,从特征空间的样本映像中提取分类相关成分,以消除冗余信息,改善样本分布。由此构建的NLCCA-LSVC集成分类器具有优良的预测性能。经模拟数据的测试,并实际用于两个复杂的化学模式识别问题,均取得令人满意的效果,也印证了算法的有效性。     

拉曼光谱法快速检测掺入梨汁的浓缩苹果汁

应用便携式拉曼光谱仪结合化学计量学技术建立了浓缩苹果汁掺入梨汁的快速检测方法.实验表明,苹果汁和梨汁的拉曼特征频率相同,给出了特征频率及其对应的分子振动模式.发现苹果汁和梨汁在866 cm-1和1 126 cm-1处的拉曼光谱有微小差别,这可能由于苹果汁和梨汁中果糖异构体含量不同所致.用支持向量分类机(SVC)中的4个核函数对19个苹果汁、梨汁样本进行有效鉴别,准确率达100%,并使用支持向量回归机(SVR)对掺入不同含量梨汁的苹果汁样本进行建模预测,大部分样品相对误差在10%以内.     

支持向量机分类和回归用于肽的QSAR研究

使用支持向量机技术对两类肽化合物体系进行了分类和回归研究,并将其系统地与K最邻近法、多元线性回归、偏最小二乘、人工神经网络进行了比较。结果表明,对于小样本、非线性问题,支持向量机具有较强的稳定性能及泛化能力,在大多数情况下能够得到优于传统方法的建模效果。对于分类问题,支持向量机对训练集和测试集都达到了100%的分类正确率;对于回归问题,支持向量机虽对训练集样本拟合效果略低于人工神经网络,但对外部测试集却表现出较强的预测能力。     

基于模糊最小二乘支持向量机的蒸发过程预测控制研究

针对氧化铝蒸发过程的多变量、非线性和大滞后特点及不同时间和空间样本数据不同的特征,提出了基于末位淘汰机制的混沌粒子群算法的综合加权模糊最小二乘支持向量机蒸发过程预测控制方法.用变异混沌粒子群算法对模型预测控制进行滚动优化,计算出最优控制序列.以某氧化铝厂蒸发过程生产数据进行实验验证分析,结果表明: 模型预测结果中相对误差小于8%的样本达到93.9%,出口浓度稳定在240 g/L附近,其控制性能得到显著改善,同时也起到了降低能耗的目的.     

加权最小二乘支持向量机改进算法及其在光谱定量分析中的应用

为克服异常训练样本对校正模型的负面影响,提出了一种加权最小二乘支持向量机(WLS-SVM)的改进算法,解决了原有算法存在的迭代收敛问题,并将其运用于光谱定量分析.实验结果表明:与原有算法相比,WLS-SVM改进算法显著增强了对异常样本的检测能力,并大幅度地提高了校正模型的稳健性.     

基于高光谱成像技术的配方烟丝组分判别

应用近红外（1 000～2 200 nm）高光谱成像技术开展了面对像素、面对样本的配方烟丝 4种组分（叶 丝、梗丝、薄片丝、膨胀丝） 的判别研究。以样本高光谱图像的所有像素点光谱数据进行面对像素的组分判 别;以样本所有像素点的平均光谱数据进行面对样本的组分判别。采用二阶导数法结合萨维茨基－戈莱平滑 （SG）滤波对光谱数据进行预处理。通过面对像素数据的主成分分析,证实了基于面对像素的高光谱数据进行 组分判别的可行性,以前5主成分建立的支持向量机模型很好地实现了叶丝与梗丝、叶丝与薄片丝的判别任 务。建立了面对样本的4组分的K近邻和支持向量机判别模型,通过连续投影算法和二阶导数法进行特征波 长选择,筛选出具有高判别准确率的波段,组分判别率达86. 97%。     

基于多特征算法选择的太赫兹时域光谱用于水稻种子掺假研究

该文提出了一种基于太赫兹时域光谱的水稻种子模式识别方法。实验以10种不同品牌混合掺假的水稻种子为样本,基于采集的样本太赫兹时域光谱数据,通过建立Relief、随机森林（RF）、支持向量机递归特征消除（SVM-RFE）和最大相关最小冗余（mRMR）模型分别对样本光谱波长进行特征选择,最后设计分类器对4种特征选择方法处理后的样本进行分类识别。结果表明,基于布谷鸟算法（CS）优化的极限学习机模型对经RF特征选择算法提取后的样本光谱数据具有最佳识别效果,其准确率可达100%,实验对于法庭科学领域内种子的掺假鉴定具有一定的借鉴意义。     

氨基酸0D-3D信息得分矢量用于人免疫缺陷病毒蛋白酶裂解位点预测及特异性分析

收集天然氨基酸的1369种0D-3D结构信息参数,经主成分分析得一组新氨基酸描述子——氨基酸0D-3D信息得分矢量,将其用于人免疫缺陷病毒蛋白酶（HIV PR）裂解位点预测,以线性判别分析与支持向量机建模预测HIV PR裂解位点．线性判别分析与支持向量机模型对646个训练集样本的自检验识别、留一法交互验证及对100个测试集样本外部验证的马休斯相关系数分别为0．879和0．911,0．849和0．901,0．822和0．846．经受试者操作特征曲线分析表明,支持向量机对HIVPR裂解位点的预测结果优于线性判别分析．研究显示,氨基酸0D-3D信息得分矢量可进一步用于HIVPR裂解位点预测．     

基于近红外光谱的果蔬脆片品质评价方法研究

本文选取苹果脆片为研究对象,结合最新果蔬脆片研究成果和支持向量机机器学习方法,以苹果脆片近红外光谱吸收值数据样本为基础,根据不同苹果样本、不同时序的红外光谱吸收值数据特征信息,产生不同的映射关系,将红外光谱吸收值映射到对应样本在对应时刻的各项品质指标(如水分含量),然后利用支持向量机(Support Vector Machine,SVM)方法对近红外光谱吸收值的数据样本进行训练,通过调整模型参数来体现不同苹果个体间的生理差异,以取得更接近于实际情况的品质指标预测效果。模拟仿真实验表明该模型的平均相关系数(R~2)达到93.81%,均方误差(MSE)为0.0057。该研究可为果蔬脆片的评价体系提供新的参考方法。     

Predicting cis/trans structure of alkene based on infrared spectra.

The application of chemometrics to analyze the information of the cis/trans structure of alkenes in infrared spectra (IR) is introduced. For data from the OMNIC IR spectral database, two feature selection methods, Fisher ratios and genetic algorithm-partial least squares (GA-PLS), and two classification methods, support vector machine (SVM) and probabilistic neural network (PNN), have been used to obtain optimization classifiers. At last, some spectra from other IR databases are used to evaluate the optimization classifiers. It has been demonstrated that both the SVM and PNN optimization classifiers could give preferable predictive results about the cis and trans structures of alkene.     

Provenance and indirect dating study on Hongzhou Kiln porcelains by INAA

Over 100 sherds were collected from Hongzhou Kiln of Jiangxi Province, China and their body samples were analyzed by INAA. The compositional differences between samples with the same period but different provenance were observed. The contents of Na, Rb, Ba, Fe, Th, U, etc., were found to be different among samples from different periods, which implies they may be used to identify the manufacturing age indirectly. The analytical data were further processed by principal components analysis (PCA). The statistical results show that all the sherds can be classified into 5 groups: (1) Eastern Han and Eastern Wu; (2) Western Jin, Eastern Jin and Southern Dynasties; (3) Sui; (4) Early Tang and Middle Tang; (5) Late Tang and Five Dynasties.     

计算机辅助对农药杀虫剂质谱信息的分类预测

对农药质谱信息的研究有助于辅助完成农药残留的鉴别和农药前体化合物的筛选。根据GB 4839-1998,选出4类杀虫剂,其常见化学结构有有机氯类、有机磷类、氨基甲酸酯类和拟除虫菊酯类农药。从NIST2.0质谱数据库中提取针对这4类结构的质谱数据,经过数学变换以及遗传算法和偏最小二乘回归相结合(GA-PLS)特征选择后,确定最优的质谱特征集,最后使用K最邻近法(KNN),支持向量机(SVM),助推法与分类回归树(AdaBoost-CART)构建预测模型。实验表明,SVM和AdaBoost-CART使用仅含有少量的特征组成的最优特征集,可以得到较好的预测结果。     

Comparison of a portable micro-X-ray fluorescence spectrometry with inductively coupled plasma atomic emission spectrometry for the ancient ceramics analysis

Two multielement instrumental methods of analysis, micro X-ray fluorescence spectrometry (micro-XRF) and inductively coupled plasma atomic emission spectrometry (ICP-AES) were applied for the analysis of 7th and 5th century B.C. ancient ceramic sherds in order to evaluate the above two methods and to assess the potential to use the current compact and portable micro-XRF instrument for the in situ analysis of ancient ceramics. The distinguishing factor of interest is that micro-XRF spectrometry offers the possibility of a nondestructive analysis, an aspect of primary importance in the compositional analysis of cultural objects. Micro-XRF measurements were performed firstly directly on the ceramic sherds with no special pretreatment apart from surface cleaning (micro-XRF on sherds) and secondly on pressed pellet disks which were prepared for each ceramic sherd (micro-XRF on pellet). For the ICP-AES determination of elements, test solutions were prepared by the application of a microwave-assisted decomposition procedure in closed high-pressure PFA vessels. Also, the standard reference material SARM 69 was used for the efficiency calibration of the micro-XRF instrument and was analysed by both methods. In order to verify the calibration, the standard reference materials NCS DC 73332 and SRM620 as well as the reference materials AWI-1 and PRI-1 were analysed by micro-XRF. Elemental concentrations determined by the three analytical procedures (ICP-AES, micro-XRF on sherds and micro-XRF on pellets) were statistically treated by correlation analysis and Student's t-test (at the 95% confidence level).     

Study on the provenance of ancient Yaozhou celadon made at Lidipo and Shangdian during Kin period using nuclear technology

Yaozhou Kiln at Lidipo and Shangdian are two independent porcelain kiln groups of Yaozhou kiln series in Shanxi Province. Both of them were consisted of some individual porcelain kilns. The samples of 20 pieces of porcelain sherds produced in Shangdian and 43 pieces of porcelain sherds made in Lidipo sites which produced in Kin Dynasty (1115–1234 A.D.) have been collected. The main chemical compositions in body were determined by X-ray fluorescence (XRF). The contents of trace elements were measured using neutron activation analysis (NAA). Principal component analysis (PCA) and stepwise discriminant analysis were used to study the provenance characteristic of these samples. The results indicated that the main components and trace elements in the specimen can be used to reveal the provenance characteristic.     

Computer-assisted prediction of pesticide substructure using mass spectra

Mass spectral classifiers of 16 substructures that are present in basic structures of pesticides have been investigated to assist pesticide residues analysis as well as screening of pesticide lead compounds. Mass spectral data are first transformed into 396 features, and then Genetic Algorithm-Partial Least Squares (GA-PLS) as a feature selection method and Support Vector Machine (SVM) as a validation method are implemented together to get an optimization feature set for each substructure. At last, a statistical method which is AdaBoost algorithm combined with Classification and Regression Tree (AdaBoost-CART) is trained to predict the 16 substructures presence/absence using the optimization mass spectral feature set. It is demonstrated that the optimum feature sets can be used to predict the 16 pesticide substructures presence/absence with mostly 85-100% in recognition success rate instead of the original 396 features.