Research On Near Infrared Spectroscopy Analysis And Application Of Flax Fiber

Bast fibers belong to natural plant fiber and is the important raw material in textile processing.It has the unique style of good moisture absorption,high strength,etc.Fabrics with bast fiber are cool,crisp and comfortable to wear.Commonly used bast fibers including ramie,flax,kenaf,jute,hemp,apocynum and so on.The appearance and chemical properties of different types of bast fibers are similar,but the market value of the final product is very different.Known as"the queen of plant fibers",flax fiber and its blended products are one of the most popular textiles in the international market.They are favored by consumers because of the characteristics of"green health care".As the main parameters of bast fibers,the contents of cellulose,hemicellulose,lignin and other chemical components have important guiding significance for evaluation,scientific research and market application of raw material in the field of bast fiber crops in China.At present,there is no unified standard for qualitative identification of bast fibers,and it is greatly affected by human factors and environmental factors.Most of the quantitative research on the chemical composition of bast fibers and the blended products of bast fibers and other fibers is still in the stage of traditional Chinese wet chemical component test.In this study,a relatively accurate and rapid NIR model prediction system was established based on the traditional wet chemical analysis method and the combination of modern NIR spectroscopy and stoichiometry.The effective identification of several different bast fibers,the quantitative prediction of the main chemical components of the flax fiber model,the flax fiber and kenaf fiber comprehensive model and the quantitative prediction of the flax content in the flax/viscose and flax/cotton blends were achieved.The main work and conclusions of this study are as follows:（1）Flax fiber was taken as the research object to determine the optimal spectral acquisition parameters such as sample size,sample filling thickness,etc.,to provide a basis for improving the prediction accuracy of the NIR model.After comparing with 30 times repeating NIR spectrum test of five different samples of the same kind of flax fiber such as 5 mm long fiber bundle,20 mesh fiber powder,40 mesh fiber powder,60 mesh fiber powder,it showed that discrete degree of spectrum（the differential absorbance value between the maximum absorbance and minimum absorbance at the same wave number）decreased with the decrease of the sample size.The minimum discrete degree of spectrum and fluctuation value of mean square deviation（SD）was obtained when the particle size was 60 mesh.After comparing with30 times repeating NIR spectrum test of different test weights（thickness）of 0.2g,0.4g,0.6g,0.8g,1.0g and 1.2g,it was found that the spectral dispersion first decreased and then increased with the increase of test weights（thickness）.When the test weight（thickness）was 1.0g,the minimum discrete degree of spectrum and fluctuation value of mean square deviation（SD）was obtained.（2）By collecting the near-infrared spectra of different bast fibers,combined with spectral pretreatment,SIMCA（Soft Independent Modeling Class Analog）recognition technology was further used on the basis of principal component analysis to effectively identify various bast fibers.The results showed that the principal component analysis models of ramie,flax,kenaf,jute,hemp and apocyanum were established,and the variances of the six models were as high as 97%when the principal component variable factor was 3,the original data information was retained to a large extent.Combined with SIMCA pattern recognition analysis,the classification results were obtained according to the data of class spacing,recognition rate and rejection rate.The maximum relative distribution distance of the six fibers is 9216.139,and the minimum is72.138.The SIMCA classification model can clearly distinguish six kinds of fibers and the order of class spacing is as follow:Flax～hemp>flax～ramie>flax～jute>flax～kenaf>flax～apocynum>jute～ramie>jute～hemp>jute～apocynum>ramie～hemp>apocynum～hemp>kenaf～jute>kenaf～apocynum>ramie～apocynum>kenaf～hemp>kenaf～ramie.In SIMCA recognition mode,100%recognition rate and 100%rejection rate were obtained for the calibration set and prediction set of six fibers,which realized fast,nondestructive,simple and accurate fiber identification.（3）Based on the traditional wet chemical analysis system of bast fibers,NIR spectroscopy together with chemometric analysis was applied to the quantitative prediction of cellulose,hemicellulose and lignin contents of 43 kinds of flax fiber.PLS quantitative analysis method was proved out to be better than PCR based on larger Rp~2and RPD and smaller RMSEP when establishing the NIR prediction model.The spectral range from 10000 to 4000 cm^-1with the FD（first derivative）and SNV（standard normal variable correction）pretreatment gained a better prediction result,with Rc~2of 0.968,Rp~2of 0.955,RMSEC of 0.777%,RMSEP of1.060%,and RPD of 4.641 for cellulose,the spectral range from10000 to 4000 cm^-1with the Baseline and FD pretreatment gained a better prediction result,with Rc~2of 0.959,Rp~2of 0.906,RMSEC of 0.511%,RMSEP of 0.678%,and RPD of 3.305 for hemicellulose,and the spectral range from 6900 to 5600 cm^-1with the FD pretreatment gained a better prediction result,with Rc~2of 0.936,Rp~2of 0.769,RMSEC of 0.351%,RMSEP of 0.455%,and RPD of 2.366 for lignin.（4）For the main components like cellulose,hemicellulose and lignin,the flax fiber and kenaf fiber comprehensive model was established to explore the organic combination of qualitative analysis and quantitative analysis of flax fiber and kenaf fiber,and then to expand the prediction ability of NIR models and improve the stability of models.The principal component analysis model of the flax fiber and kenaf fiber were built for the main components of cellulose,hemicellulose,and lignin,and the supervised SIMCA pattern recognition method was used to classify sample types.The prediction results showed that within the 5%confidence interval,the fiber of flax and kenaf in the prediction set were 100%identified.PLS method was used to establish the flax fiber and kenaf fiber comprehensive models of the main components of cellulose,hemicellulose and lignin,respectively.The spectral range from 10000 to 4000 cm^-1with the FD and SNV pretreatment gained a better prediction result,with Rc~2of 0.975,Rp~2of0.961,RMSEC of 0.961%,RMSEP of 0.952%,and RPD of 6.756 for cellulose,the spectral range from10000 to 4000 cm^-1with the Baseline and FD pretreatment gained a better prediction result,with Rc~2of 0.957,Rp~2of 0.878,RMSEC of 0.492%,RMSEP of 0.712%,and RPD of2.882 for hemicellulose,and the spectral range from 6900 to 5600 cm^-1with the FD pretreatment gained a better prediction result,with Rc~2of 0.990,Rp~2of 0.981,RMSEC of 0.616%,RMSEP of 0.788%,and RPD of 7.471 for lignin.（5）Comparing the main chemical composition model of flax fiber and the comprehensive model of the flax fiber and kenaf fiber,the results showed that the overall quality of comprehensive model of two kinds of fiber for cellulose and lignin content is better than that of a single fiber model,while the results for hemicellulose content were opposite.In the comprehensive model,the optimal spectral range for cellulose and hemicellulose is 10000-4000 cm^-1,and the optimal pretreatment mode is FD combined with SNV and Baseline combined with FD respectively.The optimal spectral range for lignin was 6900-5600 cm^-1,and the optimal pretreatment method was FD.The NIR spectra of flax fiber andαcellulose showed that the peaks and trends of their NIR spectra were consistent in the spectral range of 8000-4000 cm^-1,and the NIR spectra of flax fiber andαcellulose are slightly different in the range of 12500-8000 cm^-1 as the effect of spectral absorption of other chemical component groups,color and noise.According to the NIR spectra before and after hemicellulose removal,the absorbance of NIR spectra in the full spectrum was different,but the spectral trend was consistent.Combining the absorption range of NIR combination bands,the primary frequency doubling absorption band and the secondary frequency doubling absorption band,it is theoretically proved that the optimal modeling range of cellulose and hemicellulose is reasonable in the spectral range of 10000-4000 cm^-1.（6）The NIR spectroscopy together with chemometric analysis were applied to determine the fiber contents of flax/viscose and flax/cotton blends conveniently and accurately.A series of linen/viscose and flax/cotton samples with increasing ratio of linen（0%,10%,20%,30%,40%,50%,60%,70%,80%,90%,and 100%）were fabricated respectively.After pretreated by smoothing,baseline offset and multiplicative scattering correction,the errors of spectra caused by test conditions,sample particle size and morphology was eliminated.Seven characteristic spectral bands which were closely related to the chemical bonds of cellulose were identified:A:4000 to 4080 cm^-1,B:4227 to 4442 cm^-1,C:4658 to 4891 cm^-1,D:5110 to 5253 cm^-1,E:5556to 5701 cm^-1,F:5756 to 5936 cm^-1,G:6666 to 7067 cm^-1.For each blends,six of eleven samples with flax fiber contents of 0%,20%,40%,60%,80%,and 100%were chosen as the calibration subset,and five samples with flax fiber contents of 10%,30%,50%,70%,and 90%as the validation subset to build the MLR model and PLS model.These two high quality models have been proved to be an accurate and effective method for determining the flax contents in flax/viscose and flax/cotton blends,and the quality of PLS model is better than MLR model.Comparing with the evaluation indexes of the effect of full range and characteristic spectral range PLS models,the reduced range demonstrated better model fitness than full range model with Rp~2 by 0.998,RMSEP by 1.182%and RPD by 24.336 for flax content prediction in flax/viscose blends,with Rp~2 by 0.992,RMSEP by 2.529%and RPD by 11.239 for flax content prediction in flax/cotton blends,which indicted that the predicted flax contents were in good agreement with the reference values.Comprehensive analysis showed that,combining modern NIR spectroscopy with stoichiometric analysis technology,a systematic identification method for the properties of bast fibers can be established effectively to realize fast,and accurate fiber identification.The NIR quantitative analysis model of the flax fiber for main chemical components could be established to predict the content of cellulose,hemicellulose and lignin of flax fiber accurately,the comprehensive NIR qualitative and quantitative analysis model of flax and kenaf fiber for main chemical components could be established to achieve quickly identification of two fibers and realize accurate prediction of the content of cellulose,hemicellulose and lignin on the basis of broadening the modeling database,and the NIR quantitative analysis model of the flax fiber for flax/viscose and flax/cotton blends can be established to achieve the rapid and accurate prediction of flax content in flax/viscose or flax/cotton blends.The results lay a good foundation for the evaluation and scientific research of bast fibers.