论文部分内容阅读
用近红外光谱法对人工林杨木的苯醇抽提物质量分数进行了快速测定。用国家标准方法测定了42个杨木木材样品苯醇抽提物质量分数,并用近红外光谱仪(LabSpec Pro FR/A114260)测定相应的光谱。在350～2 500 nm、1 300～2 050 nm、2 050～2 500 nm 3个不同的光谱区域,采用未处理、Baseline、一阶导数、二阶导数等光谱的预处理方法,并采用PLS1、PLS2、PCR等3种不同的建模方法,建立了相应的校正模型与交互验证模型。结果表明:当光谱区域为2 050～2 500 nm、光谱数据进行一阶导数预处理、采用PLS1的建模方法、主成分数为7时,建立的校正模型预测效果最佳,校正模型的相关系数r=0.997 8、均方根误差为0.000 3、标准误差为0.000 3,验证模型r=0.736 0,均方根误差为0.006 1、标准误差为0.006 1;采用建立的模型对未参与建模的样本进行预测,其预测结果与实测结果之间的相关系数为0.918 6。
Near infrared spectroscopy was used to determine the mass fraction of benzene alcohol extract of poplar plantation. The content of benzene alcohol extract of 42 poplar wood samples was determined by the national standard method, and the corresponding spectra were determined by near infrared spectroscopy (LabSpec Pro FR / A114260). Three different spectral regions of 350 ~ 2 500 nm, 1 300 ~ 2 050 nm and 2 050 ~ 2 500 nm were pretreated with the spectra of untreated, Baseline, first derivative, second derivative and so on. PLS1 , PLS2, PCR and other three different modeling methods, the establishment of the corresponding calibration model and interactive verification model. The results show that when the spectral region is 2 050 ~ 2 500 nm, the spectral data is first-order derivative preprocessing, the PLS1 modeling method is adopted. When the main component is 7, the calibration model has the best prediction effect and the correlative model The coefficient r = 0.997 8, the root mean square error 0.000 3, the standard error 0.000 3, the verification model r = 0.736 0, the root mean square error 0.006 1, the standard error 0.006 1; the established model is not involved in modeling Of the sample prediction, the correlation coefficient between the prediction and the measured results was 0.9186.