近红外光谱结合SIMCA模式识别法检测木材表面节子

Rapid Detection of Knot Defect in Wood Surface by Near Infrared Spectroscopy Coupled with SIMCA Pattern Recognition

利用近红外光谱结合SIMCA模式识别法来检测马尾松木材单板节子。结果表明,通过培训集样本建立的基于主成分分析的SIMCA判别模型对有无节子两种类型样本进行回判和对未知节子类型的样本(包括无节子和有节子样本)的判别正确率均达到90%～100%,说明应用近红外光谱结合SIMCA模式识别法可以快速有效地检测木材表面的节子缺陷。

A study was performed to rapidly detect knots in Pinus massoniana veneer by near infrared （NIR） spectroscopy cou- pled with soft independent modeling of class analogy （SIMCA） pattern recognition as well as principal component analysis. The discriminant accuracy by the SIMCA model based on principal component analysis was between 90% and 100%. Resuits showed that NIR spectroscopy coupled with SIMCA pattern recognition could be used to rapidly detect knot defect in wood veneer.