Abstract Pure cashmere content (PCC) is a key parameter for assessing the quality of scoured cashmere. The traditional manual analytical method is quite laborious and time-consuming. A novel method has been developed by using near-infrared (NIR) spectroscopy combined with chemometrics and other mathematical methods. The accuracy of PCC calibration model, which is of great importance in practical application, remains a challenge due to the fact that the NIR diffuse reflectance spectra of scoured cashmere are badly influenced by absorbed moisture and light scattering. To improve the prediction accuracy, moisture elimination was first adopted to decrease the adverse effect of moisture on the spectra by subtracting the spectral contribution of water from the raw spectra. Spectral reconstruction was subsequently designed to take full advantage of the information contained in the residual values obtained after multiplicative scatter correction (MSC) pretreatment to retrieve the light scatter signal correlated with the physical properties of pure cashmere. Both the above two methods can improve the prediction performance of the partial least-squares regression (PLSR) model, and their combination can achieve the optimal model with root mean of square error of prediction (RMSEP) of 5.18 %, which satisfy the accuracy requirement for PCC measurement. The proposed method is rapid, low-cost, and environment-friendly for evaluating the quality of scoured cashmere fibers and shed light on the NIR-based analysis of the complex systems with valuable physical information.

中文翻译：

---

采用近红外光谱结合除湿和光谱重建策略的新方法快速免干测定洗净羊绒中的纯羊绒含量

摘要 纯羊绒含量（PCC）是评价精练羊绒质量的关键参数。传统的人工分析方法相当费力费时。通过使用近红外 (NIR) 光谱结合化学计量学和其他数学方法开发了一种新方法。PCC 校准模型的准确性在实际应用中具有重要意义，但由于洗涤过的羊绒的 NIR 漫反射光谱受到吸收水分和光散射的严重影响，因此其准确性仍然是一个挑战。为了提高预测精度，首先采用水分消除，通过从原始光谱中减去水的光谱贡献来减少水分对光谱的不利影响。随后设计光谱重建以充分利用乘法散射校正 (MSC) 预处理后获得的残差值中包含的信息，以检索与纯羊绒物理特性相关的光散射信号。以上两种方法都可以提高偏最小二乘回归（PLSR）模型的预测性能，它们的组合可以实现预测均方根误差（RMSEP）为5.18%的最优模型，满足精度要求用于 PCC 测量。所提出的方法快速、低成本、环保，可用于评估洗净的羊绒纤维的质量，并阐明基于 NIR 的复杂系统的分析，并具有有价值的物理信息。