近红外光谱谱线特性对物质浓度分析误差影响的研究

引用本文

赵喆, 王慧, 王慧泉, 何鑫伟, 缪竟鸿, 王金海. 近红外光谱谱线特性对物质浓度分析误差影响的研究. 光谱学与光谱分析, 2019,39(4): 1070-1074
ZHAO Zhe, WANG Hui, WANG Hui-quan, HE Xin-wei, MIAO Jing-hong, WANG Jin-hai. Influence of Spectral Characteristics on the Accuracy of Concentration Quantitatively Analysis by NIR. Spectroscopy and Spectral Analysis, 2019,39(4): 1070-1074. 复制到剪切板

Doi: 10.3964/j.issn.1000-0593(2019)04-1070-05
Permissions

光谱学与光谱分析所有

近红外光谱谱线特性对物质浓度分析误差影响的研究

赵喆^1,^2,³, 王慧¹, 王慧泉^1,^2,^3,^*, 何鑫伟¹, 缪竟鸿^1,², 王金海^1,^2,^*

1. 天津工业大学电子与信息工程学院, 天津 300387

2. 天津市光电检测技术与系统重点实验室, 天津 300387

3. 天津大学精密仪器与光电子工程学院, 天津 300072

*通讯联系人 e-mail: huiquan85@126.com

作者简介: 赵喆, 1986年生, 天津工业大学讲师 e-mail: zhaozhe@tjpu.edu.cn

收稿日期: 2018-02-04 接受日期: 2018-07-29

基金项目: 国家自然科学基金项目(61705164), 中国博士后科学基金项目第61批资助

摘要

为解决近红外光谱法分析物质浓度过程中缺乏可测度分析而导致测量过程存在一定盲目性问题, 研究在已知测量条件、样品种类、被测组分以及建模分析方法的条件下, 利用近红外光谱谱线特性作为参数, 在大量样品近红外光谱采集和标准法测得浓度数据等工作前, 对被测物质浓度的分析误差做大致估算。经过大量尝试和试验提出等效信噪比(ESNR)和谱线重叠系数(OC)两个重要参数, 其中ESNR反映待测组分吸光度占总吸光度的比重, 而OC则反映待测组分近红外光谱曲线间的重叠程度。通过理论仿真得到光谱分析中用经典的偏最小二乘回归建立定量分析模型时谱线特性与物质浓度分析误差的关系, 分别计算ESNR和OC与被测组分浓度分析误差(RMSE)的关系, 并且研究两个谱线参数的独立性。利用理论分析得到结果对浓度为8%~12%乙醇水溶液进行可测度分析, 并与近红外光谱法分析的实际结果进行比较。研究通过理论仿真得到使用光谱分析中经典的偏最小二乘回归建立定量分析模型时谱线特性与物质浓度分析误差的关系, 其中ESNR与RMSE成反比关系, 而OC与被测组分分析误差成非线性的单调关系, 并且验证了ESNR和OC两个参数的独立性。通过理论计算和乙醇水溶液近红外光谱检测实验对等效信噪比和谱线重叠系数与光谱分析浓度误差的定量关系进行讨论, 通过理论分析得到的乙醇浓度RMSE预估值为0.30%, 近红外光谱分析实际RMSE为0.32%, 相对误差6.67%, 二者结果相符。实现了在测量条件、样品种类、被测组分以及建模分析方法已知的条件下基于近红外光谱分析的待测组分含量理论误差的定量计算和实验验证。该研究明确了对近红外光谱法分析物质浓度有明确定量关系的两个谱线参数, 给出了使用光谱分析中经典的偏最小二乘回归建立定量分析模型时的分析误差经验曲线, 以及利用曲线进行近红外光谱法待测组分浓度可测度分析方法。结果表明所提出的ESNR和OC两个谱线特性参数的有效性, 以及分析误差预估方法的有效性。为近红外光谱法待测组分浓度定量分析提供了有效、快捷的预估方法, 完善了近红外光谱法成分含量可测度分析理论, 对近红外光谱法物质浓度定量分析研究具有一定指导意义。

关键词: 光谱重叠系数; 等效信噪比; 近红外光谱; 可测度分析

中图分类号:O443.4 文献标识码:A

Influence of Spectral Characteristics on the Accuracy of Concentration Quantitatively Analysis by NIR

ZHAO Zhe^1,^2,³, WANG Hui¹, WANG Hui-quan^1,^2,^3,^*, HE Xin-wei¹, MIAO Jing-hong^1,², WANG Jin-hai^1,^2,^*

1. School of Electronics and Information Engineering, Tianjin Polytechnic University, Tianjin 300387, China

2. Tianjin Key Laboratory of Optoelectronic Detection Technology and Systems, Tianjin 300387, China

3. School of Precision Instrument and Opto-Electronics Engineering, Tianjin University, Tianjin 300072, China

Abstract

In order to solve the problem of measurement blindness caused by the lack of measurable analysis in the the near-infrared spectroscopy, we can roughly estimate the analytical error of the concentration of the tested substances using the spectral characteristics of near-infrared spectroscopy under the known conditions of measurement, sample types, components under analysis and modeling and analysis methods, before a large number of samples were collected by near-infrared spectroscopy and concentration data measured by standard method. In the research, two important parameters, ESNR and OC, were proposed and tested. ESNR reflects the proportion of the component absorbance to the total absorbance, while OC reflects the overlap degree between near-infrared spectral curves of the components. We got the relationship between spectral characteristics and concentration analysis error when using the classical partial least squares regression in spectral analysis to establish quantitative analysis model through theoretical simulation. The relationship between ESNR and OC and the concentration of analyte (RMSE) was calculated respectively, and the independence of the two spectral parameters was also studied. The results of theoretical analysis were used to measure the concentration of aqueous ethanol solution between 8% and 12%, and compared with the actual results of near infrared spectroscopy. The relationship between the spectral characteristics and the concentration analysis errors when using partial least squares regression to establish a quantitative analysis model was obtained through theoretical simulation. ESNR is inversely proportional to RMSE, and OC is in a non-linear monotonic relationship with the measured component analysis error, and the independence of ESNR and OC was verified. The quantitative relationship between ESNR and OC and spectral concentration error was discussed by theoretical calculations and near-infrared spectroscopy of ethanol aqueous solution. The RMSE of ethanol concentration was 0.3% which was estimated by theoretical analysis, and the RMSE of near infrared spectroscopy was 0.32%. The relative error was 6.67%. We have realized the quantitative calculation and experimental verification of the theoretical error of the content of the tested components based on near infrared spectroscopy under the conditions of the measurement conditions, the types of samples, the components to be measured, and the methods of modeling and analysis. This study identified two spectral parameters that have a clear and quantitative relationship with the concentration of the measured component in NIR spectroscopy. The analytical accuracy empirical curve was established when using the classical partial least-squares regression in spectral analysis. In addition, the analysis of the measurable degree of the concentration of the components could also be tested by near infrared spectroscopy. The results showed the effectiveness of the ESNR and OC in this paper, as well as the analytical method of error prediction. This study provided an effective and rapid prediction method for the quantitative analysis of near infrared spectroscopy, and optimized the theory of measurable analysis of near infrared spectroscopy, which has a good guidance for the quantitative analysis of the concentration of near infrared spectroscopy.

Key words: Overlapping coefficient; Noise-signal ratio; Spectrum analysis; Measurable analysis

文章图片

引言

近红外光谱分析方法被广泛的应用于食品、医学、环境、化工等领域的检验和监测^{[1, 2, 3, 4]}, 具有无损、环保、操作便捷等优势, 但存在无特征吸收峰并且谱线重叠严重问题, 需要通过化学计量学方法建立模型。减小分析误差是光谱分析研究领域的首要问题, 现有研究主要集中在: (1) 研制更高性能的光谱采集仪器, 获得更佳的光谱数据^{[5, 6]}; (2) 使用适当的光谱预处理方法, 提高光谱数据质量^{[7, 8]}; (3) 采用合适的算法, 建立性能更优更稳定的定量分析模型^{[9, 10]}。对于光谱分析误差的影响因素研究, 主要集中在: (1) 光谱采集条件的控制和优化^[11]; (2) 光谱数据处理和建模算法研究^{[12, 13, 14]}。然而, 对于样品中待测组分本身的光谱特性对分析误差的影响目前仍然停留在定性分析。研究者们普遍认同: 光谱数据的信噪比越高, 分析误差越小; 待测组分所占比重越大, 分析误差越小; 待测组分谱线与背景谱线重叠程度越小, 分析误差越小。但是, 这些谱线特性与分析误差的定量关系研究鲜有报道。

现有近红外光谱分析研究存在一定的盲目性, 缺乏研究中的可测度分析环节, 通常是先采集大量数据, 然后尝试各种建模方法。因此, 我们对谱线特性在近红外光谱分析误差的影响进行系统的分析, 提出等效信噪比和谱线重叠系数两个参数, 通过理论仿真给出光谱分析中经典的偏最小二乘回归定量分析模型的误差经验曲线, 并进行乙醇水溶液浓度分析验证。

1 理论分析

1.1 等效信噪比

在实际检测中通常对待测组分的相对分析误差进行评估, 因此引入了等效信噪比这一参数。若仪器获得光谱数据的信噪比SNR, 组分k的等效信噪比ESNR, 有

$ESNR = \frac{A_{k}}{A} SNR$ (1)

其中A为样品总吸收光谱, A_k为组分k的吸收光谱。进而研究在相同数据处理和分析方法条件下, 定量分析均方根误差RMSE(建模集均方根误差RMSEC和预测集均方根误差RMSEP的均值作为RMSE)与等效信噪比ESNR的关系。

1.2 重叠系数

经过多种尝试和计算, 提出重叠系数(overlapping coefficient, OC), 此参数与RMSE成单调关系。重叠系数OC的定义如下:

若把样品看作由待测组分k和非待测组分的背景B这两种组分构成, 将两组分消光系数曲线ε _k(λ )和ε _B(λ )面积归一化, 即令ε '_k(λ )= $\frac{ε_{k} (λ)}{\sum_{λ}^{=} ε_{k} (λ)}$ , ε '_B(λ )= $\frac{ε_{B} (λ)}{\overset{N}{\sum_{λ = 1}} ε_{B} (λ)}$ , 使得 $\overset{N}{\sum_{λ = 1}}$ ε '_k(λ )= $\overset{N}{\sum_{λ = 1}}$ ε '_B(λ )=1, 其中N为光谱波长数。待测组分k的重叠系数OC定义为

$OC = 1 - \frac{1}{2} \overset{N}{\sum_{λ = 1}} | ε'_{k} (λ) - ε'_{B} (λ) |$ (2)

其几何意义为: 面积归一化后, 待测组分k与背景B的消光系数曲线重叠部分面积与背景组分曲线所围面积的比值。由定义可知, 重叠系数OC的取值范围为[0, 1], 其值越接近1, 说明待测组分k与背景B的消光系数曲线重叠程度越大。

2 实验部分

在ESNR和OC两个光谱曲线参数固定时, 正弦波、三角波、高斯曲线等不同谱线形状对分析结果无显著影响。考虑到近红外光谱形状与多高斯曲线叠加的形状相似, 仅列出通过改变峰宽和峰距的高斯型光谱曲线获得不同谱线特性参数OC对近红外光谱分析误差影响的实验、结果及分析。

2.1 等效信噪比对近红外光谱分析误差影响研究

首先在谱线重叠系数OC及其他参数固定的情况下, 构造ESNR的单变量实验。共512个样本, 设定吸收光谱数据长度为512个波长, 信噪比为1 000, 组成光谱矩阵S。各样品中待测组分浓度c_k为均值为1%的0.8%~1.2%均匀分布随机数, 背景组分浓度为100%-c_k。构造三组消光系数曲线, 第1组和第2组幅值相同而形状不同, 第3组幅值为前两组的1/10, 计算得到待测组分k的等效信噪比ESNR为: 6.3~221.2。

建模集和预测集按比例7∶ 3在样品集中随机选取。使用偏最小二乘回归方法建立c_k定量分析模型, 主因子数由逐一交叉验证法确定为2, 记录三组消光系数所对应的待测组分的均方根误差RMSE。

2.2 重叠系数对近红外光谱分析误差影响研究

在ESNR固定的条件下, 通过设定被测组分和背景组分消光系数曲线吸收峰的位置、高度和宽度等来改变两曲线的重叠程度, 从而探究OC对定量分析误差的影响。

用高斯曲线构造1 000个等效信噪比为500的吸收光谱作为样品。根据朗伯比尔定律, 设定光程长为10 mm, 随机抽取70%的样品作为建模集, 其余为预测集。使用偏最小二乘法建立光谱预测模型, 经过逐一交叉认证选取主因子数为2。通过改变两个高斯曲线的峰距比较16个OC时的RMSE。

2.3 等效信噪比与重叠系数对近红外光谱分析误差影响研究

为研究ESNR与OC对近红外光谱分析误差共同影响, 研究在2.2中实验基础上加入ESNR为50和5 000两组实验, 进行偏最小二乘建模分析, 研究OC对分析误差的影响。若RMSE按比例变化, 则可将ESNR和OC视为两个独立变量。

2.4 乙醇水溶液近红外光谱分析误差影响研究

为验证通过ESNR和OC对近红外光谱分析误差影响研究的实用性, 对乙醇水溶液样品进行近红外光谱采集及分析。使用Ocean Optics公司NIRQUEST512近红外光谱仪对乙醇和纯水的近红外光谱进行测量, 积分时间8 ms, 样品池厚度10 mm, 得到待测组分乙醇和蒸馏水的面积归一化消光系数曲线如图1所示, 可由式(2)计算得到它们的重叠系数OC。

	Figure Option View Download New Window
	图1 面积归一化后的消光系数曲线Fig.1 The extinction coefficient curve after area normalization

配制浓度为8%至12%的乙醇水溶液样品进行近红外光谱数据采集和分析, 以浓度间隔为0.2%配制21个样品构成建模集, 以0.5%为浓度间隔9个样品构成预测集。用偏最小二乘法建立乙醇浓度分析模型。

3 结果与讨论

3.1 等效信噪比对近红外光谱分析误差影响研究

在不同ESNR下, 使用偏最小二乘回归方法进行光谱定量分析, 得到待测组分k的等效信噪比ESNR与定量分析均方根误差RMSE的关系如图2所示。从图2中可以看出, RMSE与ESNR成反比关系。

	Figure Option View Download New Window
	图2 等效信噪比对浓度预测误差的影响Fig.2 The effect of equivalent SNR on concentration prediction error

3.2 谱线重叠系数对近红外光谱分析误差影响研究

ESNR为500时不同重叠系数情况下建模和预测均方误差由图3中第一组实验数据表示。当信噪比一定时, 待测组分浓度的RMSE随两组分OC的变化显著增长, 可以看出OC是决定待测组分近红外定量分析误差的重要因素。

由于近红外光谱分析中谱线重叠程度较为严重, 一般情况下OC值较大, 而第一组实验中OC为0.8~1区间的各点距离较大, 故加入第二组实验, 在保持ESNR等实验条件和建模分析方法一致的情况下, 通过改变谱线峰距和峰宽, 增加了48个OC较大时的数据点, 分析结果如图3所示。

	Figure Option View Download New Window
	图3 重叠系数对分析误差的影响Fig.3 The effect of OC on analytical error

由此, 通过上述实验及分析, 得到近红外光谱分析中, 被测组分和背景组分的重叠程度可以用光谱重叠系数OC进行表述, 并且图3即为OC与建模分析均方根误差RMSE的关系。该关系曲线不能用简单函数进行拟合, 实际使用时可使用分段插值等方式进行计算。

3.3 等效信噪比与重叠系数对近红外光谱分析误差影响研究

ESNR比为50, 500和5 000的样品数据进行建模分析后RMSE按ESNR比倍数进行数值统一, 分别乘以100, 10和1。所得结果如图4所示。

	Figure Option View Download New Window
	图4 不同ESNR情况下OC对分析误差的影响Fig.4 The effect of OC on analytical error in different ESNR

通过对不同ESNR的分析结果进行对比, 可以得到在不同ESNR条件下下, 被测组分浓度RMSE随OC单调变化且变化趋势是一致的, 且各组RMSE的数值上相差的倍数与ESNR的倍数一致。由此, 通过研究得到OC与ESNR比对近红外光谱定量分析误差的影响是相对独立的。均方根误差RMSE可以表示为ESNR和OC的函数, 即

$RNSE = f (ESNR, OC)$ (5)

3.4 乙醇水溶液近红外光谱分析误差影响研究

首先通过ESNR和OC对乙醇水溶液样品近红外光谱分析误差进行预估, 使用NIRQUEST512光谱仪得到近红外光谱的信噪比约为230。依据式(1)由乙醇和纯水的消光系数曲线计算得到当被测样品中乙醇浓度均值为10%时, 被测样品中乙醇的ESNR约为20。依据式(2)计算得到乙醇与纯水的OC为0.81, 代入图5曲线可得, 当ESNR为500时, RMSE为1.20× 10^-4。按照ESNR的比值进行换算, 本实验条件下10%的乙醇水溶液的浓度分析理论上的RMSE=1.20× 10^-4× 500/20=0.30%。

	Figure Option View Download New Window
	图5 预测集乙醇水溶液浓度分析结果Fig.5 The result of the concentration analysis of the prediction set of ethanol aqueous solution

通过对本实验中配制的30例乙醇浓度为8 %至12 %的乙醇水溶液样品(21例建模, 9例预测)进行近红外光谱分析, 得到预测集的乙醇的预测浓度和真实浓度的相关性分析结果如图5所示。

预测集乙醇水溶液的浓度均方根误差RMSEP=0.32%, R²=0.98。该建模分析结果与通过本研究中提出利用ESNR和OC计算得到的RMSE预估值相符, 验证了本方法的正确性。

4 结论

ESNR和OC是在对多种谱线参数(如谱线形状、信噪比、相关系数、重叠面积等)的尝试和验证基础上提出两个有效参数。 ESNR不同于以往讨论的吸收光谱的信噪比, 而是考虑待测组分的吸光度占总吸光度的比重, 从而直接反映待测组分的信噪比。 OC这一参数的提出, 使得近红外光谱曲线之间的重叠程度有了可度量的标准。

基于ESNR和OC对近红外光谱分析误差影响的系列实验, 提出了对待测组分分析误差的预估方法。若预估值满足测量需要, 即可进行大量数据采集建立分析模型; 若不满足, 可以通过提高光谱检测仪器信噪比、改进建模分析方法或者改用除近红外光谱分析外的其他生化检测方法对待测组分浓度进行定量分析。文中关系曲线是在使用光谱分析中经典的偏最小二乘回归建立定量分析模型条件下得到的, 供本领域的研究者参考, 若使用其他建模分析方法, 需要重新计算关系曲线。

针对目前近红外光谱分析中缺乏可测度分析而普遍存在一定未知性和盲目性的问题, 通过对谱线特性的系统分析, 提出ESNR和OC两个参数, 结合理论仿真和乙醇水溶液浓度分析实验并对两个参数对光谱分析误差的影响进行讨论。为近红外光谱定量分析的研究提供了有效快捷的预估分析误差方法, 完善了近红外光谱法可测度分析理论, 对近红外光谱法组分浓度定量分析的顺利进行有较高的实际意义。

The authors have declared that no competing interests exist.

参考文献

文献选项

[1]	FAN Rui, SUN Xiao-kai, CHEN Jie, et al(范睿, 孙晓凯, 陈杰, 等). Modern Food Science & Technology(现代食品科技), 2017, 33(11): 264. [本文引用:1]
[2]	Han G, Han T, Xu K, et al. Journal of Biomedical Optics, 2017, 22(7): 77001. [本文引用:1]
[3]	CHEN Hong-yan, ZHAO Geng-xing, ZHANG Xiao-hui, et al(陈红艳, 赵庚星, 张晓辉, 等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2014, 30(8): 91. [本文引用:1]
[4]	Camara A B F, Carvalho L S D, Camilo L M M, et al. Fuel, 2017. [本文引用:1]
[5]	Feng X M, Yu H X, Yi X Q, et al. Analytical Methods, 2017, 9: 2578. [本文引用:1]
[6]	Zhao M, Downey G, O’Donnell C P. Food Control, 2016, 68: 260. [本文引用:1]
[7]	Dotto A C, Dalmolin R S D, Caten A T, et al. Geoderma, 2018, 314: 262. [本文引用:1]
[8]	Dotto A C, Dalmolin R S D, Grunwald S, et al. Soil & Tillage Research, 2017, 172: 59. [本文引用:1]
[9]	WANG Li-jie, YANG Yu-yi(王丽杰, 杨羽翼). Acta Optica Sinica(光学学报), 2017, 37(10): 350. [本文引用:1]
[10]	Holland J K, Kemsley E K, Wilson R H. Journal of the Science of Food & Agriculture, 2015, 76(2): 263. [本文引用:1]
[11]	SHEN Fei, YING Yi-bin, LI Bo-bin(沈飞, 应义斌, 李博斌). Food Science(食品科学), 2014, 35(23): 25. [本文引用:1]
[12]	WANG Hai-xia, SUO Tong-chuan, YU He-shui, et al(王海霞, 所同川, 余河水, 等). China Journal of Chinese Materia Medica(中国中药杂志), 2016, 41(19): 3537. [本文引用:1]
[13]	ZHANG Jin, CAI Wen-sheng, SHAO Xue-guang(张进, 蔡文生, 邵学广). Progress in Chemistry(化学进展), 2017, 29(8): 902. [本文引用:1]
[14]	XU Ling, LI Wei-hua, YANG Ying, et al(徐玲, 李卫华, 杨英, 等). China Environmental Science(中国环境科学), 2016, 36(5): 1426. [本文引用:1]

2017

0.0

... 引言近红外光谱分析方法被广泛的应用于食品、医学、环境、化工等领域的检验和监测^[1,2,3,4], 具有无损、环保、操作便捷等优势, 但存在无特征吸收峰并且谱线重叠严重问题, 需要通过化学计量学方法建立模型 ...

2017

0.0

2014

0.0

CHEN

Hong-yan

, ZHAO

Geng-xing

, ZHANG

Xiao-hui

, et al(陈红艳, 赵庚星, 张晓辉, 等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2014, 30(8): 91.

Abstract: Soil moisture is a key issue in using spectrum analysis method to predict soil nutrients content. The purpose of this article is to explore a method of removing the effect of soil moisture and improving the hyperspectra estimation precision of soil organic matter (SOM) content. Firstly the soil samples were collected from agricultural fields of the brown soil in Daiyue county and the cinnamon soil in Huantai county, Shandong province, China. The hyperspectra of the moisture and sieved dry soil samples were measured using the ASD FieldSpec 3 and transformed to the first deviation. Because the soil moisture content and its coefficient of variation (CV) of the brown soil samples was relatively high, the brown soil samples were divided into two groups, additionally the all brown soil samples and the cinnamon soil samples, here there were four-group soil samples. Secondly, based on the difference between the moisture and dry spectra, the characteristic spectra of soil moisture were selected by singular value decomposition (SVD) in combination with correlation analysis, then the correcting coefficients of removing moisture factor from soil hyperspectra were built to reconstruct the corrected spectra of the wet samples. Finally the estimation models of the soil organic matter content were built using the partial least squares (PLS) regression based on the uncorrected and corrected spectra of the wet samples. The results indicated that using singular value decomposition to correct the moisture spectra could partly reduce the correlation coefficients between the soil moisture content and the hyperspectra in most range of spectra, and for the four-group soil samples including two for each brown soil grouped by the soil moisture content gradient, all brown soil and cinnamon soil, the coefficient of determination (R2) and relative prediction deviation (RPD) of models based on the corrected spectra were improved signally with the calibration R2 of 0.85, 0.82, 0.74 and 0.76 (an increase of 0.02-0.09), and the calibration root mean squares error (RMSE) of 0.19%, 0.20%, 0.23% and 0.19% (reduce of 0.01% - 0.03%), the validation R2 of 0.78, 0.77, 0.72, and 0.76 (an increase of 0.06 and 0.15) the validation RMSE of 0.21%, 0.15%, 0.21% and 0.15% (reduce of 0.01% - 0.08%) except for the cinnamon soil samples (increased 0.02 percentage), the validation RPD of 2.03, 2.02, 1.86 and 1.98 (an increase of 0.17 - 0.43), especially for the three-group samples with the smaller CV in soil moisture content. The models reached the good performance from needing improvement and could achieve better prediction accuracy of the soil organic matter content. Therefore, the experiments indicated that the method was effective to remove the soil moisture influence and improve the prediction accuracy of the soil organic matter content from hyperspectra. In addition, in order to achieve the better prediction accuracy, the soil samples should be grouped by the soil moisture content to reduce the dispersion degree of the soil moisture.

土壤水分的影响是当前采用光谱分析法预测土壤养分含量的关键问题，该文旨在探索去除土壤水分影响、提高有机质高光谱定量估测精度的方法。首先采用地物光谱仪进行湿土和过筛干土的高光谱测试，并进行一阶导数变换；然后，采用奇异值分解（singular value decomposition，SVD）结合相关分析筛选土壤水分特征光谱，构建去除水分因素的修正系数，形成湿土光谱的校正光谱；最后基于校正前后湿土光谱，应用偏最小二乘（partial least squares，PLS）回归构建土壤有机质含量的估测模型，并对模型进行验证和比较，分析评价校正前后光谱的预测精度。结果显示：按土壤水分含量梯度划分的2组和全部棕壤及褐土土样共4组样本校正后建模决定系数和均方根误差分别为0.85、0.82、0.74、0.76和0.19%、0.20%、0.23%、0.19%，决定系数提高了0.02～0.09，均方根误差降低了0.01～0.03百分点，验证决定系数、均方根误差和相对分析误差分别为0.78、0.77、0.72、0.76，0.21%、0.15%、0.21%、0.15%和2.03、2.02、1.86、1.98，决定系数提高了0.06～0.15，均方根误差除褐土土样提高0.02百分点外，其他样本组降低了0.01～0.08百分点，相对分析误差提高了0.17～0.43，模型决定系数和相对分析误差得到显著提升；尤其对于土壤水分含量变异系数较小的3组土样，模型从待改进级别提高到性能良好级别，对土壤有机质含量具有较好的预测准确性。说明该方法用于去除土壤水分因素影响和提高有机质含量高光谱估测精度的有效性。

2017

0.0

2017

0.0

... 减小分析误差是光谱分析研究领域的首要问题, 现有研究主要集中在: (1) 研制更高性能的光谱采集仪器, 获得更佳的光谱数据^[5,6] ...

2016

0.0

... 减小分析误差是光谱分析研究领域的首要问题, 现有研究主要集中在: (1) 研制更高性能的光谱采集仪器, 获得更佳的光谱数据^[5,6] ...

2018

0.0

... (2) 使用适当的光谱预处理方法, 提高光谱数据质量^[7,8] ...

2017

0.0

... (2) 使用适当的光谱预处理方法, 提高光谱数据质量^[7,8] ...

2017

0.0

... (3) 采用合适的算法, 建立性能更优更稳定的定量分析模型^[9,10] ...

2015

0.0

... (3) 采用合适的算法, 建立性能更优更稳定的定量分析模型^[9,10] ...

2014

0.0

SHEN

Fei

, YING

Yi-bin

, LI

Bo-bin

(沈飞, 应义斌, 李博斌). Food Science(食品科学), 2014, 35(23): 25.

In order to evaluate the influence of temperature on near-infrared spectroscopy (NIRS) analysis models of Chinese rice wine quality, NIR spectra of rice wine samples were collected at 5, 10, 15, 20, 25, 30 and 35 ℃, respectively. Calibration models for the quantification of alcohol and total sugar contents in rice wine were developed by partial least square (PLS) algorithm. The results indicated that the spectra of samples were affected by temperature. An obvious clustering tendency was detected by principal component analysis (PCA) between samples at different temperatures. The prediction ability of the calibration models was also influenced by temperature, but showing no consistent trend with temperature. The performance of the multi-temperature calibration models was suitable for practical application with low relative error values. These results provided a theoretical basis for rapid detection of rice wine quality as well as temperature correction by NIR spectroscopy.

1. College of Food Science and Engineering, Nanjing University of Finance and Economics, Nanjing 210023, China; 2. College of Biosystems Engineering and Food Science, Zhejiang University, Hangzhou 310058, China; 3. National Center for Quality Supervision and Testing of Rice Wine, Shaoxing 312071, China

为研究温度对黄酒品质近红外光谱分析模型的影响，分别在5、10、15、20、25、30、35 ℃ 7 个温度条件下采集黄酒样品的可见-近红外光谱，采用偏最小二乘法建立各温度下黄酒酒精度和总糖含量定量分析模型。结果表明：温度对样品光谱存在影响，主成分分析表明不同温度下的样品有明显聚类趋势。模型精度受温度影响较大，但并无随温度变化的一致趋势。建立的混合温度模型预测相对误差较小，有实际应用潜力。

... 对于光谱分析误差的影响因素研究, 主要集中在: (1) 光谱采集条件的控制和优化^[11] ...

2016

0.0

... (2) 光谱数据处理和建模算法研究^[12,13,14] ...

2017

0.0

... (2) 光谱数据处理和建模算法研究^[12,13,14] ...

2016

0.0

... (2) 光谱数据处理和建模算法研究^[12,13,14] ...