葡萄酒酿造过程酒精度检测方法研究

酒精度作为葡萄酒中的一项极其重要的指标，对葡萄酒的品质起着至关重要的作用。该实验以葡萄酒作为研究对象，引入红外光谱分析技术对葡萄酒的酒精度进行间接、无损的检测研究。实验中对比分析了不同红外光谱预处理方法以及建模方法在建立葡萄酒酒精度预测模型时的优劣，发现采用多元散色校正和标准归一化的去噪效果最优，然后采用主成分分析法建立的模型最优，预测相关系数（R）达0.942，标正标准差（RMSEC）、预测标准差（RMSEP）、相对分析误差（RPD）分别为0.154、0.149、2.96。最终对建立的模型进行验证，发现采用红外光谱分析技术能够快速且较准确的对葡萄酒酒精度进行在线预测，为葡萄酒的酿造过程实时监测酒精度提供了一种新方法。

Determination method of alcohol content during wine fermentation

Alcohol content is a very important index in wine, which plays an important role in wine quality. The experiment using wine as research object, the alcohol content was determined indirectly and nondestructively by infrared spectral technology. The advantages and disadvantages of different infrared spectral preprocessing methods and modeling method in wine alcohol content prediction was compared and analyzed. Results showed that multiplicative scatter correction and single nucleotide variant were the optimal methods for denoising, and principal component analysis was the optimal method for model establishment, the correlation coefficient R was 0.942, and the RMSEC, RMSEP, RPD was 0.154, 0.149, 2.96, respectively. Finally through the validation tests on the model, the results showed that the infrared spectral technology could online predict wine alcohol content rapidly and accurately, which could provide a new method for wine brewing process of real-time monitoring of alcohol content.