Study On Rapid Measurement Method For Pulpwood Characteristics Using The Near-infrared Techniques

The characteristics of the pulpwood are closely related to the quality of the terminal products,so it is necessary to analysis the characteristics of pulpwood in the production process and adjust the pulping process parameters in time according to the species and characteristics of pulpwood.However,the traditional analysis methods can not meet the need of rapid analysis in the process of industrialized continuous production.In order to reduce the cost of enterprises,improve the stability of production operation and market competitiveness,based on near-infrared spectroscopy and chemometrics system,the study developed prediction models for the characteristics of pulpwood and develop the research of rapid analysis technology.The study selected 13 kinds of common pulpwood including Eucalyptus urophylla × grandis,Eucalyptus urophylla U6,Eucalyptus urophylla L11,Eucalyptus globules,Acacia mangium,Acacia crassicarpa,Acacia aulacocarpa,Acacia auriculiformis,A.auriculiformis × A.mangium,Triploid populus tomentosa,Populus deltoids,Populus euramevicana,Populus euramericana.The holographic grating spectrometer was used as the analytical instrument.First of all,the parameters of near-infrared measured environment and spectral acquisition were determined.The content of holocellulose in the aged Eucalyptus globules was studied,and the best parameters of the milled wood prediction model was as follows: temperature between 15³0℃,relative humidity between 30%⁵0%,spectrometer scanning speed is 360°/min,spectrum collection frequency is 20 times/circle,each sample farce 3 times to take the average spectrum,wood powder is compacted by 1.41 k Pa weight pressure.The basic density of the aged Eucalyptus globules was studied and the best parameters of the wood chips model was as follows: the temperature is between 15³0℃ and the relative humidity in 30%⁸0%,spectrometer scanning speed is 360°/min,spectrum collection frequency is 20 times/circle,each sample farce 10 times to take the average spectrum.The content of chemical components and basic density of 13 kinds of pulpwood were determined,and the actual moisture content was determined under the condition of artificial control of water content.The near-infrared spectrum of various samples were collected,and the near-infrared spectrum were pretreated by various combination pretreatment methods in Matlab 8.0,and the models were established for various preprocessing methods by loading partial least squares,LASSO method,support vector machine method and artificial neural network method.Then using genetic algorithm to select the bands,comparing the selected optimal bands with those of the 3rd overtone interval,the 2nd overtone interval,the 1st overtone and combination interval and the total spectral bands respectively.The accuracy of models were improved and the performance of the models were optimized.This paper attempted to analyze the characteristic absorption of the holocellulose,Klason lignin,pentosan,cold and hot water soluble extractives,benzene-alcohol extractives,1% Na OH extractives and moisture in the optimal band from the component structure,and theoretically explained the prediction ability of the models.According to the situation of pulping industry,two kinds of mixed pulpwood of Poplar-Eucalyptus and Eucalyptus-Acacia were simulated.The near-infrared analysis models of mixing degree and chemical composition content were established,and the correlative parameters were determined.The conclusions are as follows:In the analysis of holocellulose content,the pretreatment method of smoothing,vector normalization,first derivative was adopted,1331.0²362.1 nm band was selected,and LASSO algorithm was used to establish the calibration model.The RMSEP value of the single pulpwood model is 0.48%,and the model is suitable for quality control.The RMSEP value of Poplar-Eucalyptus mixed pulpwood model is 0.6%,and the model is suitable for imprecise measurement.The RMSEP value of Eucalyptus-Acacia mixed pulpwood model is 0.57%,and the model is suitable for quality control.In the analysis of Klason lignin content,the pretreatment method of smoothing,MSC and second derivative of original spectrum was adopted,the band of 1135.4¹874.8nm and 2132.9²426.1nm was selected,and support vector machine algorithm was used to establish the calibration model.The RMSEP value of single pulpwood model is 0.42%,the model is suitable for quality control.The RMSEP value of Poplar-Eucalyptus mixed pulpwood model is 0.57%,and the model is also suitable for quality control.The RMSEP value of Eucalyptus-Acacia mixed pulpwood model is 0.47%,and the model is suitable for imprecise measurement.In the analysis of pentosan content,the pretreatment method of smoothing,vector normalization and second derivative was adopted,the band of 1328.3²342.5nm was selected,partial least squares algorithm was used to establish the calibration model.The RMSEP value of single pulpwood model is 0.72%.The RMSEP value of Poplar-Eucalyptus mixed pulpwood model was 0.81%,and the RMSEP value of Eucalyptus-Acacia mixed pulpwood model was 0.79%.The model is suitable for quality control.In the analysis of the content of cold-water-soluble extractives,the pretreatment method of smoothing,MSC and second derivative was adopted,the band of 1331.2²328.5nm was selected,the LASSO algorithm was used to establish the calibration model.The RMSEP value of single pulpwood model is 0.24%,the model is suitable for quality control.The RMSEP value of Poplar-Eucalyptus mixed pulpwood model was 0.26%,and the RMSEP value of Eucalyptus-Acacia mixed pulpwood model was 0.29%.Two mixed pulpwood models are suitable for imprecise measurement.In the analysis of the content of hot-water-soluble extractives,the pretreatment method of smoothing,MSC and second derivative was adopted,the band of 1335.6²334.7nm was selected,the algorithm of artificial neural network was used to establish the calibration model.The RMSEP value of single pulpwood model is 0.28%,the model is suitable for quality control.The RMSEP value of Poplar-Eucalyptus mixed pulpwood model is 0.29%,and the model is suitable for quality control.The RMSEP value of Eucalyptus-Acacia mixed pulpwood model is 0.32%,and the model is suitable for imprecise measurement.In order to analyze the content of benzene-alcohol extractives,the pretreatment method of smoothing,MSC and first derivative was adopted,whole-band was selected,and the partial least squares algorithm was used to establish the calibration model.The RMSEP value of single pulpwood model is 0.23%.The RMSEP value of Poplar-Eucalyptus mixed pulpwood model is 0.26%,and the RMSEP value of Eucalyptus-Acacia mixed pulpwood model is 0.29%.All three models are suitable for quality control.In 1% Na OH extractives content analysis,the pretreatment method of smoothing,vector normalization,first derivative was adopted,the band of 1139.1²362.9nm was selected,and the LASSO algorithm was used to establish the calibration model.The RMSEP value of single pulpwood model is 0.38%.The RMSEP value of Poplar-Eucalyptus mixed pulpwood model is 0.39%.The RMSEP value of Eucalyptus-Acacia mixed pulpwood model is 0.41%.All three models are suitable for quality control.In the moisture content analysis,pretreatment method of smoothing,normalized,first derivative was adopted,the band of 1346.7¹477.4nm and 1790.0¹983.3nm was selected,and the partial least squares algorithm was used to establish the calibration model.The RMSEP value of the model is 1.19%.The model has good adaptability.In the basic density analysis,the pretreatment method of smoothing,vector normalization and first derivative was adopted,whole-band was selected and the LASSO algorithm was chosen as the modeling method.The RMSEP value of the model is 6.31 kg·m^-3.The model has good adaptability.In the mixture degree analysis,the pretreatment method of smoothing,vector normalization and first derivative was adopted,the LASSO algorithm was chosen as the modeling method,and the whole-band was selected.The RMSEP value of Poplar-Eucalyptus mixed pulpwood model is 1.84%.The RMSEP value of Eucalyptus-Acacia mixed pulpwood model is 1.93%.Two models have good adaptability.According to the above conclusions,the pretreatment system of pulpwood,which can be used for grinding,fast drying,sieving and shaping,was developed,and the data analysis and control system was assembled.The three components were used to make the rapid measurement platform of the pulpwood.The rapid analysis of pulpwood was realized.