基于气相色谱-主成分分析的红木分类识别方法研究

Study on the classification and recognition of mahogany based on gas chromatograph-principle component analysis

木材化学分类法的研究较少。通过木材的化学成分和化学计量学方法,从分子的角度鉴别珍贵木材,具有重要意义。本文通过GC-FID实验,采集阔叶黄檀等5种18批次的红木样本的色谱数据,所建立实验方法重现性好。将所得色谱数据,进行色谱峰对齐和自标度化预处理,然后PCA投影。12个建模样本被分成4类,与各样本已知的植物学分类一致。以所建立的分类方法(即PCA投影空间),识别其余6个待鉴别样本,结果准确达到分离聚类。本方法利用现代分析仪器和模式识别法实现了对红木的分类和识别,为珍贵木材的化学分类鉴别法提供理论依据。

Chemical method is not well established in wood classification study. Using chromatogram data, we developed a pattern recognition method on Mahogany based on Principle Component Analysis （PCA） in this study. We collected chromatogram data from 5 Mahogany species by means of GC-FID technology with good reproducibility. After data preprocessing, such as peak alignment and antoscaling, classical chemometric method of PCA is applied to resolve the data. 12 samples from 4 species are classified into corresponding 4 groups, which is consistent with the botanic taxonomy. Then we take prediction test with our model for another 6 presumed undetermined samples. The undetermined samples are projected into the existing linear space, as a result, 3 groups fall into the corresponding known PCA clusters and 1 group is outside the PCA clusters, which is also consistent with the botanic taxonomy. The result indicates that our methods do good performance in Mahogany classification and species prediction. Our methods also provide a theoretical basis for the further research of precious wood＇s classification and discrimination.