热风干燥海带含水量近红外光谱预测模型建立及品质评价

Establishing a Near-Infrared Spectrum Prediction Model of Water Content of Hot Air Dried Kelp and Its Quality Evaluation

为实现普通热风干燥海带含水量的实时监测,采用近红外光谱技术结合偏最小二乘法建立预测模型。并用多元散射校正进行光谱预处理,采用连续投影算法提取特征波长以减小计算量。选取干燥过程中8个时间点各15个海带样品(共120个),从中随机抽取80个样品为样本集和内部验证集建模,另外40个样品为预测集对模型进行评价。结果表明,在全波长近红外光谱拟合和特征波长所建的模型中,2个模型的预测集相关系数分别达到了0.9629和0.9574,RMSEC分别为0.0651和0.0686。2个模型水分含量预测值与实际测量值吻合度均较高,且SPA-PLS模型较全波长扫描模型相比,计算量大幅降低,更适用于海带干燥过程水分的快速测定。此外,文章对干燥后的海带样品进行品质评价,分别测定干燥前后抗氧化能力、叶绿素、碘、甘露醇和总糖含量。结果表明,热风干燥法对于海带品质影响较小,对于以上活性物质有较好的保留。综上所述,所建立的近红外光谱预测模型可以实现海带干燥过程中水分含量的快速、实时检测。

In order to realize real-time monitoring of water content of kelp dried by ordinary hot air,a prediction model was established by near infrared spectroscopy combined with partial least square method.Multiple scattering correction was used to preprocess the spectrum,and continuous projection algorithm was used to extract the characteristic wavelength to reduce the computation.Fifteen kelp samples were selected at 8 time points during the drying process(120 in total),80 samples were randomly selected as sample sets and internal validation sets for modeling,and the other 40 samples were used as prediction sets to evaluate the model.The results showed that the correlation coefficient of prediction sets of the two models was 0.9629 and 0.9574,RMSEC is 0.0651 and 0.0686,respectively,in the model built by full-wavelength NIR spectral fitting and characteristic wavelength.The actual moisture content measurements were consistent with the predicted values of the two models.Since SPAPLS model was more appropriate for the rapid determination of the moisture content in the kelp during the drying process since its calculation duration was shorter than that of the full-wavelength scanning model.This study assessed the quality of the dried kelp samples,while determining the antioxidant capacity,as well as the total sugar,iodine,mannitol,and chlorophyll content before and after the drying process.According to the results,hot air drying had no significant impact on the kelp quality and was more successful in preserving the active substances.Therefore,the near-infrared spectroscopy model developed in this research was capable of rapidly detecting the moisture content level in real-time during the drying process of the kelp.