近红外光谱结合化学计量学的常见中国蜂蜜掺杂糖浆鉴别

目前我国蜂蜜市场掺假现象严重,研究一种快速、准确的方法用于市场流通领域掺假蜂蜜的鉴别具有重要的现实意义。采用近红外光谱（NIR）结合化学计量学方法对常见的天然蜂蜜以及掺假（掺杂常见糖浆）蜂蜜进行建模识别,并比较偏最小二乘-判别分析（PLS-DA）及支持向量机（SVM）对糖浆掺假蜂蜜鉴别模型的影响。首先,采集来自中国10个省份、20种常见蜂蜜的112个天然纯蜂蜜样品,以及6种常见糖浆样品按不同糖浆含量（10%,20%,30%,40%,50%,60%）配制的112个掺假蜂蜜样品,共计224个样品;通过近红外光仪器扫描获得所有样品的近红外光谱数据（波长范围400~2 500 nm）;然后,分别采用一阶导数（FD)、二阶导数（SD）、多元散射校正（MSC)、标准正态变化(SNVT)四种方式对原始光谱进行预处理;再结合PLS-DA和SVM建立天然蜂蜜和糖浆掺假蜂蜜的鉴别模型,比较不同预处理方法对两种不同建模算法建立的蜂蜜掺假鉴别模型效果。其中SVM算法的惩罚参数c和核函数参数g通过网格搜索法（GS）、遗传算法（GA）、粒子群算法（PSO）三种寻优算法进行优化。分析结果表明：光谱数据进行预处理后所建立的模型准确率均有明显提升,而对于SVM模型,惩罚参数c和核函数参数g对模型准确率的提升效果要比光谱预处理带来的提升效果更明显。在PLS-DA算法中,经FD光谱预处理后建立的模型效果最好,最佳PLS-DA模型准确率为87.50%;在SVM算法中,经MSC预处理后,再通过GS寻优,获得惩罚参数c为3.0314,核函数参数g为0.3298的条件下所建立的模型效果最好,最佳SVM模型准确率为94.64%。由此可见,非线性的SVM算法结合NIR光谱数据所建立的天然蜂蜜与糖浆掺假蜂蜜鉴别模型要优于线性的PLS-DA模型,同时表明NIR光谱结合化学计量学方法对常见糖浆掺杂的中国蜂蜜鉴别是可行的。     

利用近红外及中红外融合技术对小麦产地和烘干程度的同时鉴别

小麦是制作馒头的主要原料之一,小麦中水、蛋白质、淀粉会因产地以及烘干程度的差异而不同,进而影响到加工成馒头的品质。所以实现对小麦产地和烘干程度的快速鉴别就显得尤为重要。感官评定是鉴别小麦产地和烘干程度常用的方法,对比感官评定,光谱分析可以识别样品中的分子结构等信息。基于此,尝试利用近红外和中红外光谱融合技术实现对不同产地和不同烘干程度的小麦同时鉴别。首先选取了两个不同产地的小麦,再利用微波干燥法对两个不同产地的小麦做烘干预处理,使烘干的小麦水含量为12%±0.5%,原麦水含量为18%±0.5%。分别标记为原麦A,烘干A,原麦B,烘干B,再将小麦研磨成粉末,过100目筛网筛选后,置于自封袋中备用。随后分别采集四种小麦样品的近红外和中红外光谱信息,在Matlab 7.10的环境下使用标准正态变量变换(standard normal variable transformation, SNVT)对采集到的原始光谱数据进行预处理,利用主成分分析对预处理后的数据进行降维处理,再结合线性判别分析（linear discriminant analysis,LDA）和支持向量机（support vector machine, SVM）分别建立小麦近红外、中红外光谱数据识别模型。另外利用联合区间偏最小二乘法（synergy interval partial least square, SiPLS）筛选出利用标准正态变量变换(SNVT)预处理后的小麦近红外和中红外光谱数据特征光谱区间,将筛选出的近红外和中红外光谱数据特征光谱区间融合后再结合线性判别分析（LDA）和支持向量机（SVM）建立小麦融合光谱信息的识别模型。然后比较同种光谱数据下利用线性判别分析（LDA）和支持向量机（SVM）建立的小麦识别模型识别率、比较同种建模方法下近红外和中红外光谱数据建立小麦识别模型识别率、比较同种建模方法下光谱数据融合和单一光谱数据建立小麦识别模型识别率。结果表明,同种光谱分析方法,利用SVM建立的四种小麦识别模型识别率高于利用LDA建立的小麦识别模型识别率。同种建模方法,近红外光谱数据建立的小麦识别模型识别率优于中红外光谱数据建立的小麦识别模型识别率。而在同种建模方法下,利用SiPLS筛选出近红外和中红外光谱数据的特征光谱区间数据融合后建立小麦识别模型识别率最高,光谱数据融合后结合LDA建立的小麦识别模型校正集识别率为98.75%,预测集识别率为97.50%;而将此选择的变量结合SVM建立的小麦识别模型的校正集和预测集识别率都达到100.0%。对比利用单一光谱数据建立的小麦识别模型识别率,光谱数据融合之后建立的小麦识别模型识别率得到显著提高,该研究从纵向和横向上全面地比较了光谱数据建立的小麦模型识别率,结果可为更准确地运用光谱融合技术建立小麦产地以及烘干程度识别模型提供参考。     

Three-class classification models of logS and logP derived by using GA–CG–SVM approach

In this investigation, three-class classification models of aqueous solubility (logS) and lipophilicity (logP) have been developed by using a support vector machine (SVM) method combined with a genetic algorithm (GA) for feature selection and a conjugate gradient method (CG) for parameter optimization. A 5-fold cross-validation and an independent test set method were used to evaluate the SVM classification models. For logS, the overall prediction accuracy is 87.1% for training set and 90.0% for test set. For logP, the overall prediction accuracy is 81.0% for training set and 82.0% for test set. In general, for both logS and logP, the prediction accuracies of three-class models are slightly lower by several percent than those of two-class models. A comparison between the performance of GA–CG–SVM models and that of GA–SVM models shows that the SVM parameter optimization has a significant impact on the quality of SVM classification model. Electronic supplementary material  The online version of this article (doi:) contains supplementary material, which is available to authorized users. Hui Zhang and Ming-Li Xiang are contributed equally.     

Local linear regression for data with AR errors

In many statistical applications, data are collected over time, and they are likely correlated. In this paper, we investigate how to incorporate the correlation information into the local linear regression. Under the assumption that the error process is an auto-regressive process, a new estimation procedure is proposed for the nonparametric regression by using local linear regression method and the profile least squares techniques. We further propose the SCAD penalized profile least squares method to determine the order of auto-regressive process. Extensive Monte Carlo simulation studies are conducted to examine the finite sample performance of the proposed procedure, and to compare the performance of the proposed procedures with the existing one. From our empirical studies, the newly proposed procedures can dramatically improve the accuracy of naive local linear regression with working-independent error structure. We illustrate the proposed methodology by an analysis of real data set.     

流形判别分析和支持向量机的恒星光谱数据自动分类方法

尽管经典的分类方法支持向量机SVM在天文学领域广泛应用,但其只考虑类间的绝对间隔而忽略类内的分布性状,因而分类性能有待于进一步提升。鉴于此,提出一种新颖的基于流形判别分析和支持向量机的恒星光谱数据自动分类方法。该方法引入流形判别分析的两个重要概念：基于流形的类内离散度M_W和基于流形的类间离散度M_B。所提方法找到的分类面同时保证M_W最小且M_B最大。可建立相应最优化问题,然后将原最优化问题转化为QP对偶形式求得支持向量和判别函数,最后利用判别函数判断测试样本的类属。该方法的最大优势在于进行分类决策时,不仅考虑样本的类间信息和分布特征,而且还保持了各类的局部流形结构。SDSS恒星光谱数据上的比较实验表明该方法的有效性。     

竞争性自适应重加权算法和相关系数法提取特征波长检测番茄叶片真菌病害

基于竞争性自适应重加权算法(CARS)和相关系数法(CA)特征波长选择方法,提出了利用可见-近红外高光谱成像技术检测番茄叶片灰霉病的方法。首先获取380～1 023 nm波段范围内80个染病和80个健康番茄叶片的高光谱图像,然后提取染病和健康叶片感兴趣区域(ROI)的光谱反射率值,作为番茄叶片灰霉病鉴别模型的输入来建立支持向量机(SVM)鉴别模型,训练集和验证集的鉴别率都是100%。研究进一步通过CARS和CA提取特征波长,分别得到5个(554, 694, 696, 738和880 nm)和4个(527, 555, 571和633 nm)特征波长,然后分别建立CARS-SVM和CA-SVM鉴别模型。结果显示,CARS-SVM模型中训练集和验证集的鉴别率都是100%,CA-SVM模型中训练集和验证集的鉴别率分别是91.59%和92.45%。以上结果说明了从可见-近红外高光谱图像中提取的光谱反射率值用于检测番茄叶片的灰霉病是可行的。     

光谱特征波长的SPA选取和基于SVM的玉米颗粒霉变程度定性判别

利用波长范围在833～2 500 nm的傅里叶变换近红外光谱(Fourier transform near infrared spectroscopy, FT-NIR)对不同霉变程度的玉米颗粒进行检测区分。首先,为避免光谱数据首尾噪声影响,对比四种常见的预处理方法,最终选择移动平均平滑法对原始光谱数据进行预处理;然后为选出合适的样本集划分方法以提高模型预测性能,对常见的四种方法进行对比,最终利用SPXY(sample set partitioning based on joint x-y distance)法进行样本集划分;进一步为减少数据量,降低维度,使用连续投影算法(successive projections algorithm, SPA)提取出7个特征波长,分别为833,927,1 208,1 337,1 454,1 861和2 280 nm;最后,将七个特征波长数据作为输入,选取径向基函数(radial basis function, RBF)作为支持向量机(support vector machine, SVM)核函数,取参数C=7 760 469,γ=0.017 003建立判别模型。SVM模型对训练集和测试集的预测准确率分别达到97.78%和93.33%。另取不同品种的玉米颗粒,以同样的标准挑选样品组成独立验证集,所建立的判别模型对独立验证集的预测准确率达到91.11%。结果表明基于SPA和SVM能有效地对玉米颗粒霉变程度进行判别,所选取的7个特征波长为实现在线霉变玉米颗粒近红外检测提供了理论依据。     

An intelligent approach for engine fault diagnosis based on Hilbert–Huang transform and support vector machine

Based on the techniques of Hilbert–Huang transform (HHT) and support vector machine (SVM), a noise-based intelligent method for engine fault diagnosis (EFD), so-called HHT–SVM model, is developed in this paper. The noises of a sample engine under normal and several fault states are first measured and denoised by using the wavelet packet threshold method to initially lower the noise level with negligible signal distortion. To extract fault features of the engine, then, the HHT is selected and applied to the measured noise signals. A nine-dimensional vector, which consists of seven intrinsic mode functions (IMFs) from the empirical mode decomposition (EMD), maximum value of HHT marginal spectrum and its corresponding frequency component, is specified to represent each engine fault feature. Finally, an optimal SVM model is established and trained for engine failure classification by using the fault feature vectors of the noise signals. Cross-validation results show that the proposed noise-based HHT–SVM method is accurate and effective for engine fault diagnosis. Due to outstanding time–frequency characteristics and pattern recognition capacity of the HHT and SVM, the newly proposed HHT–SVM can be used to deal with both the stationary and nonstationary signals, and even the transient ones. In the view of applications, the HHT–SVM technique may be suggested not only to detect the abnormal states of vehicle engines, but also to be extended to other fields for failure diagnosis in engineering.