基于近红外光谱技术的杉木总酚含量模型构建与应用

Establishment and application of total phenols content model of Chinese fir based on near-infrared spectroscopy

  • 摘要: 【目的】木样总酚含量化学测定耗时长、过程复杂,建立杉木木样总酚含量的快速无损检测模型,对实现木材无损检测及木材腐朽预测具有重要意义。【方法】试验以114个杉木(Cunninghamia lanceolata)木样为研究对象,用福林酚法测定样品总酚含量,利用MPA傅立叶变换光谱仪对杉木木材进行漫反射光谱数据采集。将木样分为校正集和验证集,通过不同光谱预处理方法和建模方法建立总酚的定量模型,选择出最优模型并用验证集对其进行验证。【结果】测定的114个杉木木样中总酚含量变异幅度大,可用于构建近红外模型。对114个杉木木样进行近红外光谱扫描,得出建模光谱范围为9403.9~7498.4 cm-1、6102.1~5446.4 cm-1及4605.5~4242.9 cm-1。对杉木木样的近红外光谱进行预处理,得出最优组合:标准正态变量转换法(SNV)和一阶导数,采用偏最小二乘回归法(PLS)建立模型最优。校正集和交叉验证集的决定系数分别是0.8679和0.7549;校正均方根误差(RMSEE)和交叉验证均方根误差(RMSECV)分别为0.448和0.586,数值均较小且接近,说明模型具有很好稳定性;预测均方根误差(RMSEP)和相对标准偏差(RPD)分别为0.521和2.16,说明模型可进行定量分析。【结论】采用近红外光谱技术检测杉木总酚含量可行,能为木材化学成分快速测定提供一种有效、无损方法。受拟合规则影响,构建的模型虽然不能用于精确定量测定,仍可应用于日常科研和生产检测,在木材材质预测及良种选育等方面具有广阔应用前景。

     

    Abstract: 【Objective】Chemical determination of total phenol content in wood samples was time-consuming and complex. It was of great significance to establish a rapid non-destructive testing model for total phenol content in Chinese fir wood samples for non-destructive testing and wood decay prediction.【Method】In the experiment, 114 Chinese fir (Cunninghamia lanceolata) wood samples were taken as the research objects, and the total phenol content of the samples was determined by Folin phenol method. The MPA Fourier transform spectrometer was used to collect the diffuse reflectance spectrum data of Chinese fir wood. The wood samples were divided into correction set and verification set. Quantitative models of total phenol were established by different spectral pretreatment methods and modeling methods, and the optimal model was selected and verified by validation set.【Result】The variation range of total phenol content was large, which could be used to construct the near-infrared model. 114 Chinese fir wood samples were scanned by near-infrared spectroscopy, and the modeling spectral range was 9403.9-7498.4 cm-1, 6102.1-5446.4 cm-1 and 4605.5-4242.9 cm-1. The near-infrared spectra of Chinese fir wood samples were pretreated, and the optimal combination was obtained:standard normal variable transformation(SNV) and first derivative, partial least squares regression(PLS) was used to establish the model. The determination coefficients of calibration set and cross validation set of total phenol content in Chinese fir wood samples were 0.8679 and 0.7549, respectively. The root mean square error of calibration(RMSEE) and root mean square error of cross validation(RMSECV) were 0.448 and 0.586, respectively, the values were small and close, indicating that the model had good stability. The root mean square error of prediction(RMSEP) and relative standard deviation (RPD) were 0.521 and 2.16, respectively, indicating that the model could be used for quantitative analysis.【Conclusion】It is feasible to detect the total phenol content of Chinese fir by near-infrared spectroscopy, which provides an effective and nondestructive method for the rapid determination of wood chemical components. Affected by the fitting rules, although the constructed model can not be used for accurate quantitative determination, it can still be used in daily scientific research and production detection. It has broad application prospects in wood material prediction and improved variety breeding.

     

/

返回文章
返回