Prediction Of The Main Physical And Mechanical Properties Of Eucalyptus Pellita Wood By Near Infrared Spectroscopy

The thesis took wood of Eucalyptus pellita as the research object based on the principle of near infrared spectroscopy (NIR). Air-dry density, modulus of elasticity (MOE), bending strength (MOR) and comprehensive strength parallel to grain of E.pellita were measured according to the national standard "Test methods for physical and mechanical properties of wood." Besides the air-dry density and modulus of elasticity of E.pellita increment cores and slice specimens were measured by SilviScan-3 and the rapid prediction of wood physical and mechanical properties of small clear samples of E.pellita by near infrared spectroscopy was researched by NIR in detail. Also, the impact of the thickness, moisture content and surface roughness of samples as well as different methods of measured air-dry density on the prediction results of wood properties by NIR were studied. Main results were concluded as following:(1) Analyzing the model which built using different wavelength, different pretreated NIR spectra and different sections, it choosed the whole wavelength, second derivative pretreated NIR spectra and average NIR spectra from three sections to build the air-dry density with partial least-squares. The correlation coefficient between prediction values and actual measured values was 0.87, SEP was 0.041, RPD was 2.00. It concluded that the NIR analysis technique can realize the rapid and convenient prediction for the air-dry density of wood samples of E.pellita.(2) Analyzing the rapid prediction of wood mechanical properties of small clear samples of Eucalyptus pellita by NIR. Choosing whole wavelength, second derivative pretreated NIR spectra and NIR spectra from radial and tangential sections to build calibration. The results showed that the relationships were good between the laboratory-determined mechanical properties values and NIR prediction values of mechanical properties, whose correlation coefficients of prediction model ranging from 0.89 for modulus of elasticity through 0.92 for bending strength to 0.90 for comprehensive strength parallel to grain of wood, SEP were from 1.505,8.621 and 3.496 respectively, the ratio of performance to deviation were 2.12,2.63 and 2.40. It concluded that the NIR analysis technique can realize the rapid and convenient prediction for wood mechanical properties of E.pellita.(3) Thickness, moisture content and surface roughness of samples and different methods of measured air-dry density had a great impact on prediction results of wood properties by NIR. The models of 2mm thickness and 5mm thickness were better than the models of 10mm thickness and 20mm thichness. In terms of RPD all four models are above 2.40, which demonstrated that NIR analysis technique can predict the air-dry density of wood samples of E.pellita, but the sample thickness affect the forecast results severely; differences results of samples of the model exist in different roughness, so the same roughness should be maintained as far as possible for modelling; also sample moisture content can not be ignored which was an important factor, the model was the best when the samples moisture content was 12%. under a certain moisture content collected near-infrared spectroscopy to built the model to predict wood air-dry density was best; it can realize the rapid and convenient prediction for the air-dry density of wood samples of E.pellita slice specimens using X-ray densitometer, SilviScan combined with NIR, the correlation coefficient between NIR prediction values and X-ray densitometer measured values is 0.95, while the correlation coefficient between NIR prediction values and SilviScan measured values is 0.93.(4) Determining the air-dry density and longitudinal elastic modulus of increment cores samples and slice specimens samples based on SilviScan-3, and building the calibrations combined with NIR. It is analized that the correlation coefficient between prediction values and actual measured values was good, It demonstrates that the NIR analysis technique can realize the rapid and convenient prediction for air-dry density and longitudinal elastic modulus of increment cores samples and slice specimens samples of E.pellita. Meanwhile, analyzing the correlation air-dry density and longitudinal elastic modulus between increment cores samples and slice specimens samples, it showed that the correlation were good between them, increment cores can be instead of slice specimens to measure air-dry density and longitudinal elastic modulus of standing tree and evaluated the wood properties of standing tree rapidly.