Abstract The aim of this study was to develop applications of near infrared spectroscopy (NIRS) to detect certain anomalies in cork used in the manufacture of stoppers for wine bottling. 326 spectra of cork planks were used in a qualitative analysis to evaluate the feasibility of NIRS to discriminate between pure cork, yellow stain and lignified cork. Classification and validation errors varied between 0 % and 11.4 %. In addition, 360 spectra of cork granulate were used in a quantitative analysis to determine the percentage of yellow stain and corkback. A root mean standard error of cross validation (RMSECV) of 1.50 % was obtained for yellow stain percentage (R² = 99.62 %). For corkback percentage RMSECV was 2.96 % (R2 = 98.53 %). These results show that NIRS technology has potential, alongside traditional methods, to determine the quality of cork used to manufacture stoppers.

中文翻译：

---

使用近红外光谱 (NIRS) 对软木塞异常进行定性和定量评估

摘要 本研究的目的是开发近红外光谱 (NIRS) 应用，以检测用于制造葡萄酒瓶塞的软木塞中的某些异常。326 个软木板光谱用于定性分析，以评估 NIRS 区分纯软木、黄色染色和木质化软木的可行性。分类和验证错误在 0 % 和 11.4 % 之间变化。此外，软木颗粒的 360 度光谱用于定量分析，以确定黄色染色和软木塞的百分比。对于黄色染色百分比 (R² = 99.62 %)，获得了 1.50 % 的交叉验证根平均标准误差 (RMSECV)。对于软木塞百分比，RMSECV 为 2.96 % (R2 = 98.53 %)。这些结果表明 NIRS 技术与传统方法一样具有潜力，