废旧纺织品天然纤维成分近红外光谱分析模型

The textile industry is of great significance to the healthy development of China's light industry system. Textile raw materials in China are very shortage. But a great deal of waste textiles are not used, and improper handling of waste textiles causes environmental pollution. Sorting recycling of waste textiles can save a lot of textile raw materials, and reduce the consumption of the resources of natural fibers. Recycling of waste textiles at present mainly by means of manual sorting. This way is low efficiency and high cost. It is difficult to meet the needs of large-scale finely sorting and grading of waste textiles. This project studies the classification problem of waste textiles by means of near infrared spectra analysis. The near infrared spectra qualitative analysis models of natural fiber components of waste textiles will be constructed by combining the multi model method and Support Vector Machine method. The waste textiles will mainly be divided into three categories by the models. For the two categories that contain natural fiber components, the near infrared spectra quantitative analysis models that have a certain universality will be constructed respectively by the multi model method. The models will be used to determine the contents of natural fiber components of these two categories. This will lay a foundation for finely sorting and grading of waste textiles. Because the use of non manual sorting recycling of waste textiles has a very good application prospect, and the researches on establishing near infrared spectra analysis models by multi model methods is relatively less both at home and abroad, the research of this project has important significance.

纺织品产业对我国轻工业体系健康发展意义重大。我国纺织原材料异常紧缺，但大量废旧纺织品却没能得到利用，对废旧纺织品的不当处理还造成了环境污染。分类回收废旧纺织品可大量节省纺织原材料，减少对天然纤维资源的消耗。目前回收废旧纺织品主要采用人工分拣方式。这种方式效率低、成本高，难以满足对废旧纺织品进行大规模精细分拣、分级的需要。本项目用近红外光谱分析方法研究废旧纺织品的分类问题。将多模型方法与支持向量机方法相结合建立废旧纺织品天然纤维成分近红外光谱定性分析模型，用模型将废旧纺织品主要划分为三个大类。对含有天然纤维成分的两个大类，分别用多模型方法建立具有一定普适性的近红外光谱定量分析模型，用于测定其天然纤维成分含量，为废旧纺织品的精细分拣、分级奠定基础。由于采用非人工分拣方式回收废旧纺织品具有很好的应用前景，且国内外使用多模型方法建立近红外光谱分析模型的研究相对较少，所以本项目的研究具有重要意义。

用基于主成分分析的支持向量机（简称PCA-SVM）方法,分别建立了废旧纺织品纤维素成分近红外光谱定性分析模型和蛋白质成分近红外光谱定性分析模型。模型具有很好的分辨能力。综合使用上述两个模型，将废旧纺织品样本分成了三类 ：纤维素基类、蛋白质基类、纯合成纤维类。.将多模型方法与PCA-SVM方法成功结合形成了新的定性分析方法。用上述方法建立了废旧纺织品的近红外光谱定性分析模型。所建模型的分类效果好于用PCA-SVM方法所建模型的分类效果，且模型的稳定性明显得到提高。该研究结果为废旧纺织品的大规模精细分拣和分级奠定了一定的基础。.用多模型方法建立了废旧纺织品纤维素含量的近红外光谱分析模型，然后用校正集里纤维素含量高于70%的样本建立了适用于高纤维素含量样本的预测模型，用校正集里纤维素含量低于30%的样本建立了适用于低纤维素含量样本的预测模型。由此减小了对纤维素含量较高或较低的样本的预测误差。将上述结果与原来模型的预测结果相结合，纤维素含量最终预测值所对应的拟合优度和相关系数分别比原来有所提升。上述方法有望用于建立其它材料某些化学成分的近红外光谱定量分析模型。.将多模型方法与偏最小二乘法相结合用于建立废旧纺织品蛋白质含量近红外光谱分析模型。结果表明：模型的预测效果好于单独使用偏最小二乘法所建模型的预测效果。如果废旧纺织品样本涉及多种蛋白质，对每种蛋白质分别建立定量分析模型的预测效果好于不对蛋白质进行分类所建模型的预测效果。