基于近红外光谱的卷烟质量投影模型

摘要

Projection model of cigarette quality based on NIR spectra was established to improve stability of cigarette formula and quantification of cigarette quality. Principal component analysis was used to reduce dimensionality of NIR data, based on which, Fisher discriminant functions were established for projection of cigarette products. Results showed that projective images could exactly describe the quality of different cigarette brands. Application of this model can facilitate the evaluation of cigarette quality.%为提高卷烟配方的稳定性，科学量化卷烟的质量水平，建立基于卷烟产品近红外光谱信息的质量投影模型。通过主成分分析对光谱数据进行降维，利用主成分建立Fisher判别函数对各卷烟产品进行投影。实验结果表明，模型得到的二维投影图能直观反映各卷烟牌号质量的高低关系，为卷烟质量的评价提供了客观的科学依据。