Study On Qualities Analysis Of Cashmere Raw Material Based On IR Spectroscopy And FESEM Technology

Cashmere and wool fiber detection is the most fundamental but also the weakest link of the cashmere and wool industry in our country.The technique level of detection is far from meeting the developing demands of cashmere and wool industry and exists a large gap when compare to England,Japan,Germany,Italy,and Australia:cashmere and wool automation and continuous degree are not high,especially in the detection of raw material of cashmere and wool,our country is far behind the developed countries.Because of lack of rational detection methods,the qualities of most of domestic cashmere and wool cannot meet the requirement of process and this has become development bottleneck of cashmere and wool industry.On the one hand, Cashmere has its own advantages and the features of scarce production and high price.On the other hand,it relates to the benefits of consumers and merchants,and more important it relates to state reputation in the import and export trade,the identification and content measurement of cashmere is extremely important.In foreign countries,a series of detection and technique technical standards had been applied in order to protect their own self-equity and avoid presenting of adulterate and inferior products.China is a large country for production of cashmere and also is a major country for import and export of finished or semi-finished products of cashmere,it is necessary to reinforce the detection and supervision in this field to make our country play a due role in the area of quality estimation of cashmere,and further promote the improvements of cashmere and wool raw materials and developments of cashmere and wool products,ensure our country's precious cashmere resource to develop toward the direction of Scientific,orderly,and high added value.Considering the problem and deficiency of current technologies in the information acquisition system and the actual situation of our country,cashmere and wool qualities and varieties and fiber morphology information rapid acquisition were studied in this thesis.The main research work and achievement were as follows:1.Using near infrared spectroscopy technology to detect the yield and moisture content of cashmere.Stepwise regression analysis and back-propagation neural network were applied to build linear and nonlinear regression model.R~2,SEP,RMSEP were used to evaluate the model.This method has so many advantages as rapid,nondestructive,and no pollution compared with traditional detection methods.2.Wavelet analysis method was used for de-noising of infrared spectroscopy signals.Comparative analysis of the effects of three threshold de-noised models were given,the using of wavelet de-noised methods had improved model deviation caused by external interference,and made the model more robust and anti-jamming.3.Based on the infrared spectroscopy characteristics of protein,ash content,fat content of cashmere and wool materials,long wave near and middle infrared spectroscopy diffused pattern were applied to measure the detection index above mentioned.Pattern recognition method of BP neural network,projection pursuit regression,least squares support vector regression were applied to build data analysis model.The results illustrate BP neural network is the most stable and universial for the three quality parameters spectral analysis.4.JASCO FT/IR - 4100 infrared spectrometer were applied to study the relationship between long wave near and middle infrared spectroscopy and protein,ash content,fat content of cashmere and wool materials,spectral transmission pattern were applied to measure the detection index above mentioned.Pattern recognition method of BP neural network,Projection Pursuit Regression,least squares support vector regression was applied to build data analysis model.The results illustrate infrared spectroscopy in transmission pattern is better than that in diffuse pattern on detection and model building of protein,ash content,fat content of cashmere under the test conditions described in this paper.5.Applying Vis/near infrared spectroscopy technique to identification of origins of cashmere,principal component analysis and support vector machine were used to analyze the spectroscopy of different varieties of cashmere.One against all support vector classification was used to discriminate the origins of cashmere.Several kernel functions were compared to find the best model.The results illustrate that SVM muti-class classition model with Gaussian kernel function has a good identification effect,and can work as a new method on rapid identification of cashmere material varieties.6.SIRION field-emission scanning electron microscope was applied to obtain microscopic image of cashmere and wool fibers.The distinctions between the cashmere and wool were studied based on geometric shape parameters.The new concept of scale cover bilateral was proposed to enhance the identification capabilities to a great extent.7.Texture analysis of gray level co-occurrence matrices was presented to analyze the SEM image of cashmere and wool fibers.22 texture feature parameters such as energy,contrast, correlation,homogeneity,entropy and so on were extracted from image,identification model based on the texture features were built.The result illustrate:texture analysis cannot been affected by grayscale intensity and directivity,it can extracted features have not correlation with rotation.Application of gray level co-occurrence matrices make the problem of cashmere and wool fiber identification better solved.8.Image process and analysis were used to the SEM image of cashmere and wool fibers. The region properties distribute operators that reflect the scale shape characteristics were applied to build the identification model of cashmere and wool.The results illustrate scale shape characteristics can reflects the morphologies destination between cashmere and wool to some degree,but the resolution cannot meet the demand,so it is necessary to optimize analysis method of microscopic image of the fiber scale.Conclusively,on the one hand,macroscopic detection of the index such as yield,moisture content protein,ash content,fat content of cashmere and wool materials were proposed by using infrared spectral techniques;on the other hand,microscopic analysis of the morphologies characteristics such as geometric shape of fiber,texture feature and region properties of the scale of cashmere and wool were studied based on field-emission scanning electron microscope image. The relatively new data analysis methods such as stepwise regression analysis,back-propagation neural network,wavelet de-noised method,projection pursuit regression,least squares support vector regression were used to built the quantitative relation between IR spectroscopy and the features index of cashmere and wool,morphology analysis methods of image were applied to build the qualitative destination between cashmere and wool fiber.This research provides advanced and objective detecting methods and theoretical foundations in the field of qualities control in production and transaction of cashmere and wool materials.