The Near Infrared Spectral Analysis Mathematical Model Of Populus Tomentosa And Acacia

In this paper, we use some chemical experimental data and near infrared spectral data of populus tomentosa samples and acacia confusa samples, take the multiple model method as the core tool, establish the near infrared spectral analysis model of populus tomentosas’alpha cellulose content and that of populus tomentosas’lignin content,and establish the near infrared spectral analysis model of acacias’benzyl alcohol extract content.The specific results and conclusions are as follows:1. We built up the near infrared spectral prediction model of populus tomentosa’s alpha cellulose content with multiple model method, and study the effect of different pretreatment methods to the model prediction,they are the smoothing preprocessing of spectral data, the first derivative pretreatment, the second derivative pretreatment and the combination of them.Results shows that the second derivative pretreatment combined with smoothing pretreatment is the best pretreatment method, the goodness of fit is 0.9308by this method, the correlation coefficient between the predicted values and the experimental values is 0.9648.2. We built up the near infrared spectral prediction model of populus tomentosa’s alpha cellulose content and that of populus tomentosa’ slignin content with multiple model method.Then,according to the approximate linear relationship between the alpha cellulose contents and the lignin contents, using the alpha cellulose content with the smaller prediction error to help the lignin content with a larger prediction error build the new prediction model. The prediction effect of the lignin content is improved.The correlation coefficient between the predicted values and the experimental values of the lignin contents increases from 0.7473 to 0.7962, the average relative error of the predicted values of the lignin contents reduces to 4.67% from 5.59%.3. We built up the near infrared spectral analysis model of acacia rachii’s benzyl alcohol extract content and that of acaciaklassen lignin content with multiple model method.Then, using klassen lignin content with less prediction error to help benzyl alcohol extract content build a new prediction model. The prediction effect of benzyl alcohol extract content is improved.Goodness of fit of the model increases from 0.7928 to 0.8271, the correlation index between the predicted values and the experimental values increases from 0.9074 to 0.9225.But when modeling, the existence of the approximate linear relationship between the two kinds of chemical composition content is not required.4. We produced the alpha cellulose contents’ fast forecasting software based on multiple model method.In near infrared spectral analysis, multiple model method is a kind of modeling method which haven’t been widely used.So this articles’ research is meaningful.In this paper, the study of spectral data preprocessing method can provide experience for other near infrared spectral analysis mathematical modeling with multiple model method;The research about using the chemical composition content with a good prediction effect to help the chemical composition content with a usual prediction effectbuild a new prediction model, is expected to be used for some chemical composition content who’s prediction effect is general, and to improve their near infrared spectral analysis effects.