基于梨枣轻微损伤的可见/近红外光谱判别研究

利用可见/近红外光谱技术对梨枣轻微损伤的分类判别建模方法进行研究。分别采用PLS-LDA(线性)和LS-SVM(非线性)建立判别模型,分析比较不同预处理方式和建模波段对模型精度的影响。结果表明:经9点平滑预处理后的短波近红外(780~1100nm)PLS-LDA模型判别效果最佳,校正集和预测集的正确识别率分别达到97.8%和96.7%。     

注水肉无损检测技术现状与发展趋势分析

针对目前市场上注水肉现象加重问题,有必要研究注水肉的检测技术。首先对注水肉的传统检测方法进行简要论述,分析总结了传统方法的弊端;然后对现阶段肉品含水率的4种主要无损检测方法:生物电阻抗法、核磁共振方法、超声波法和近红外光谱法的研究现状及特点进行了总结分析,着重叙述了近红外光谱技术在肉品含水率及品质检测中的研究情况及技术特点,提出了未来技术发展方向。     

近红外多组分分析中异常样本识别方法

近红外光谱分析中,异常样本的存在严重影响定标模型的预测性能和适配性。基于 X / Y 联合的ODXY异常样本识别和剔除方法,提出并证明了一种专用于多组分分析的MODXY异常样本识别方法。实验采用80组玉米近红外光谱数据,利用不同异常样本识别方法剔除异常样本后建立玉米含水率、含油率、蛋白质含量和淀粉含量4种组分的偏最小二乘预测模型,采用预测均方差和决定系数作为评价指标比较所建模型的性能,检验MODXY方法在多组分分析中的异常样本识别能力。实验结果表明：在近红外多组分分析中,MODXY方法在大多数情况下具有更好的异常样本识别能力;MODXY方法和ODXY方法均有一定的适用范围,它们更适合于相应组分化学值的相对标准偏差较大的情况。     

近红外技术(NIR)在小麦商品粮收购中的应用研究

利用近红外谷物品质分析仪 (Perten DA910 0 ) ,对陕西省某县 6 2 4户农民 2 0 0 0年夏季上缴粮库的公购粮进行现场品质分析。结果认为 :人工主观评定的商品粮等级与其蛋白质含量、水分含量、硬度、沉淀值、烘烤体积无显著相关性 ;近红外谷物品质分析仪 (Perten DA910 0 )完全能够满足粮库验粮的需要 ;小麦籽粒硬度、蛋白质含量是商品粮收购定级的重要指标     

Spectrophotometrical characteristics in the near infrared region in beech (<Emphasis Type="Italic">Fagus crenata</Emphasis>) and pine (<Emphasis Type="Italic">Pinus densiflora</Emphasis>) litters at the various decomposing stages

We examined the relationships between the absorptional characteristics in the near infrared region and the chemical changes of decomposing beech (Fagus crenata) and pine (Pinus densiflora) litters. Spectra as well as the concentrations of chemical substances approached each other and converged with decomposition, although both initial characteristics differed markedly between beech and pine. This indicated that the fundamental chemical structures were almost the same, although their organochemical composition differed. Specific absorption bands for lignin, polysaccharide, and protein were identified at 2,140 and 1,670 nm, 2,270, 1,720, 1,590, and 1,216 nm, and 2,350 nm, respectively. Absorbance at 1,670 nm, peculiar band of aromatics, showed a positive correlation with lignin concentration, which suggested the relative increment of aromatics due to condensed lignin in decomposing litters. Absorbance at 2,140 nm, characterized as the C–H bond in HRC = CHR, showed a negative correlation with lignin concentration, which suggested the decrements of some structures such as side-chains in lignin polymers unrelated to aromatics. Absorbance at 2,270, 1,720, and 1,216 nm, specified to O–H/C–O/C–H bonds in saccharide, might reflect the change of polysaccharide during decomposition because they showed a positive correlation to polysaccharide concentration. In the same way, absorbance at 2,350 nm, identified to the C–H/CH₂ bonds in protein, showed a negative correlation to nitrogen concentration in decomposing litters, which might indicate that the C–H/CH₂ bonds in protein decreased with decomposition due to microbial consumption of carbon in protein. Our findings suggested the possibility that the spectral changes indicate the litter digestibility during decomposition and that also explain the compositional change in decomposing litters.     

Use of Near Infrared Spectroscopy to Measure Mechanical Properties of Solid Wood

The visible and near infrared （NIR） （350-2500 nm） spectra and the MOE of 438 small clear wood samples from Chinese fir, eucalyptus and poplar 72 were measured. Using partial least-square （PLS） modeling, the NIR spectra could be used to predict MOE and MOR of the clear-wood samples from Chinese fir and eucalyptus solid wood. NIR spectra could only be used to Predict MOE but not MOR of poplar clear-wood samples. With the exception of MoR of poplar clear-wood samples, the correlations between NIR and the mechanical properties are very strong, and the calibration and test correlation coefficients are both above 0.80.     

土壤硝态氮含量原位检测系统设计

针对现阶段土壤硝态氮测量成本较高、无法长期原位测量等问题,该研究提出了一种使用钛烧结滤芯收集土壤溶液,通过近红外光谱法检测土壤溶液中的硝酸根浓度进而得到土壤中硝态氮含量的方法,并设计了相应的检测装置。通过试验对比陶土头与钛烧结滤芯在不同土壤条件下的土壤溶液收集效果,选用钛烧结滤芯作为土壤溶液采集器收集土壤溶液,以近红外LED作为测量光源,采集人工配置土壤溶液的光谱数据,利用BP神经网络进行预训练建立硝态氮含量预测模型。建立的硝态氮含量预测模型其训练集皮尔逊相关系数、测试集皮尔逊相关系数、预测均方根误差分别达到0.997、0.995、3.43。实地测量土壤溶液并与硝酸根离子电极以及土壤养分速测仪进行对比,最大相对偏差为5.9%,可满足实际测量准确性要求。该套检测设备在深度为10～40 cm、含水率为15%以上的土壤中有较好的土壤溶液采集效果;检测装置的长期测量标准差为0.006,动态响应时间为1.4 s,具有良好的特性。试验结果表明,使用溶液吸光度数据建立的硝态氮预测模型具有较好的预测效果,可以应用于土壤溶液硝态氮原位监测,为长期自动测量土壤硝态氮及水肥一体化系统的搭建提供了一种可行的方案。     

应用NIR及主成分回归预测落叶松密度的研究

运用近红外光谱主成分回归法对落叶松样品密度进行研究，校正集的相关系数（R）为0．86，校正集标准误差（SEC）为0．01，预测集的相关系数（R）为0．89，预测集标准误差（SEP）为0．02，对未参与建模的12个未知样品进行密度预测，相关系数达0．95。研究表明，近红外光谱能够快速、准确地对落叶松样品密度进行预测，这为快速检测落叶松木材材性提供了一种新方法。     

Near‐infrared spectroscopy can predict the composition of organic matter in soil and litter

The usefulness and limitations of near‐infrared reflectance spectroscopy (NIRS) for the assessment of several soil characteristics are still not sufficiently explored. The objective of this study was to evaluate the ability of visible and near‐infrared reflectance (VIS‐NIR) spectroscopy to predict the composition of organic matter in soils and litter. Reflectance spectra of the VIS‐NIR region (400–2500 nm) were recorded for 56 soil and litter samples from agricultural and forest sites. Spectra were used to predict general and biological characteristics of the samples as well as the C composition which was measured by ¹³C‐CPMAS‐NMR spectroscopy. A modified partial least‐square method and cross‐validation were used to develop equations for the different constituents over the whole spectrum (1st to 3rd derivation). Near‐infrared spectroscopy predicted well the C : N ratios, the percentages of O‐alkyl C and alkyl C, the ratio of alkyl C to O‐alkyl C, and the sum of phenolic oxidation products: the ratios of standard deviation of the laboratory results to standard error of cross‐validation (RSC) were greater than 2, the regression coefficients (a) of a linear regression (measured against predicted values) ranged from 0.9 to 1.1, and the correlation coefficients (r) were greater than 0.9. Satisfactorily (0.8 ≤ a ≤ 1.2, r ≥ 0.8, and 1.4 ≤ RSC ≤ 2.0) assessed were the contents of C, N, and production of DOC, the percentages of carbonyl C and aromatic C and the ratio of alkyl C to aromatic C. However, the N‐mineralization rate and the microbial biomass were predicted unsatisfactorily (RSC < 1.4). The good and satisfactory predictions reported above indicate a marked usefulness of NIRS in the assessment of biological and chemical characteristics of soils and litter.     

Near-infrared spectroscopy for analysis of chemical and microbiological properties of forest soil organic horizons in a heavy-metal-polluted area

In industrial areas, heavy metals may accumulate in forest soil organic horizons, affecting soil microorganisms and causing changes in the chemical composition of the accumulated organic matter. The objectives of this study were to test the ability of near-infrared spectroscopy (NIRS) to detect heavy metal effects on the chemical composition of forest soil O horizons and to test whether NIRS may be used to quantitatively determine total and exchangeable concentrations of Zn and Pb (Zn_t, Pb_t, Zn_ex, Pb_ex) and other chemical and microbial properties in forest soil O horizons polluted with heavy metals. The samples of O horizons (n = 79) were analyzed for organic C (C_org), total N and S (N_t, S_t), Zn_t, Pb_t, Zn_ex, Pb_ex, basal respiration (BR), microbial biomass (C_mic) and C_mic-to-C_org ratio. Spectra of the samples were recorded in the Vis-NIR range (400–2,500 nm). To detect heavy-metal-induced changes in the chemical composition of O horizons principal components (PC1–PC7) based on the spectral data were regressed against Zn_t + Pb_t values. A modified partial least squares method was used to develop calibration models for prediction of various chemical and microbial properties of the samples from their spectra. Regression analysis revealed a significant relationship between PC3 and PC5 (r = −0.27 and −0.34, respectively) and Zn_t + Pb_t values, indicating an effect of heavy metal pollution on the spectral properties of the O horizons and thus on their chemical composition. For quantitative estimations, the best calibration model was obtained for C_org-to-N_t ratio (r = 0.98). The models for C_org, N_t, and microbial properties were satisfactory but less accurate. NIRS failed to accurately predict S_t, C_org-to-S_t, Zn_t, Pb_t, Zn_ex, and Pb_ex.