Preliminary Study On The Rapid Detection Of Resin Content In Wood Fiber And Particles By Near Infrared Spectroscopy

The resin content of wood-based panel could be predicted nondestructively,accurately and rapidly by Near infrared spectroscopy(NIR).It could provide a new technology to detect the resin content of wood-based panel.In this paper,the wood fiber and wood particle with various urea formaldehyde resin content ranging from 2% to 40% were used as the research objects.The models between the resin content and the near infrared spectrum of samples with different resin content were established.Based on the establishment of discrimination and prediction model,the effects of sample properties,spectral sampling methods,external factors and spectral pretreatment on the discrimination and prediction models were discussed.The methods to obtain more stable and accurate models and accurate discrimination and prediction results were discussed too.The main conclusions were as follows:(1)Thirteen kinds of wood fiber and wood particle samples with different resin content were designed and prepared.The model between the resin content and the near infrared spectrum of wood fiber and wood particle samples after spreading were established by near infrared spectroscopy.The discriminant accuracy of the model to "low,medium and high" of resin content of wood fiber and small size wood particle samples reached 80% ? 94%,and that of big size wood particle samples was 68% to 100%.The decision coefficient prediction results are from 0.84 to 0.98.The results showed using Near infrared spectroscopy could to detect the resin content of wood-based panel was feasible.But it need tobe optimized and improved.(2)The influencess of optical fiber detection distance,external light source and sample movement speed on the detection results were investiguated.The results showed that the detection distance changes had less affection on the predicted results of the wood fiber and small size wood paticles.But it had greatly impact on wood partiles with big size.To increasing the velocity of the wood particle samples to 4 m/min could reduce the accuracy of the prediction.The external light source condition did not affect the prediction results.(3)The selection of wavelength range to build NIR model had different effects on different samples.The results showed that wood fiber was suitable for the wavelength range of500 ? 1100 nm,and the discrimination accuracy was more than 70%;that of wood particle with small size was the wavelength range of 500 ? 1100 nm,and the discrimination accuracy was more than 88%;that of the wood particle with big size was the wavelength range of 500 ?2500 nm,and the discrimination accuracy was more than 80%.In order to save the research and development cost of the equipment,It was recommended to select the spectrum in the wavelength range of 500 ? 1100 nm.