光谱学与光谱分析, 2018, 38 (8): 2511, 网络出版: 2018-08-26   

基于小波分解和因子分析的白酒香型和年份鉴定的研究

Classification and Year Prediction of Chinese Liquors Based on Wavelet Decomposition and Factor Analysis
作者单位
1 江南大学理学院, 江苏 无锡 214122
2 江苏省轻工光电工程技术研究中心, 江苏 无锡 214122
摘要
提出了一种基于小波分解和因子模型分析白酒荧光光谱, 对白酒香型进行分类和年份预测的方法。 白酒的三维荧光光谱包含了其所含荧光物质信息, 对其进行小波分解, 其分解系数与特征峰的强度相关。 选取高斯小波对三维荧光光谱进行分解, 可以避免对二维荧光光谱进行分解时需要选取特定激发波长的问题。 对样品的三维荧光光谱进行小波分解后, 选取第4层近似系数构建正交因子模型, 通过因子载荷系数对白酒进行鉴别。 结果指出, 贡献率较小的因子蕴含着样品的独特信息, 在相似样品的比较中, 不容忽视。 在对10个品牌的白酒进行香型分类时, 先将样品的三维荧光光谱进行高斯小波分解, 使用第4层近似系数进行因子分析, 得到贡献率由大到小的多个因子。 根据因子的载荷系数, 对样品进行聚类分析。 结果表明, 加入贡献率较小的因子可以将正确率提高至90%。 通过对因子载荷系数与年份的相关性分析得出, 贡献率排在前六位的因子和白酒年份关系较大, 而排在后面的因子和白酒年份的相关性较小, 因此可以选取前六位的因子建立白酒年份预测模型。 通过选取不同贡献率的因子对白酒年份进行预测, 其平均误差可降低至0.9年。
Abstract
In this paper, a method to identify the flavor and year of Chinese liquors was proposed based on continuous wavelet decomposition and factor model on analyzing the fluorescence spectra of liquor. The three-dimensional fluorescence spectrum of liquor contained the information of the fluorescent substance, and its decomposition factor was related to the intensity of the characteristic peak. The decomposition of the three-dimensional fluorescence spectrum by Gaussian wavelet can avoid the problem of selecting the specific excitation wavelength when decomposing the two-dimensional fluorescence spectrum. After the wavelet decomposition of the three-dimensional fluorescence spectrum of the sample, the orthogonal factor model was constructed by the fourth layer approximation coefficient, and the liquor was discriminated by the factor loading. The results showed that the factors with small contribution contain unique information of the sample, which can not be neglected in the comparison of similar samples. In the classification of liquor flavor, the three-dimensional fluorescence spectra of the samples were decomposed by Gaussian wavelet, and the fourth-layer approximation coefficients were used for factor analysis to obtain multiple factors with large and small contribution rates. According to the factor of the factor loading, the cluster analysis was carried out. The results showed that the factor with a small contribution rate can increase the correct rate to 90%. By analyzing the correlation between the factor loading and the year of liquors, the contribution rate of the first six factors was larger than that of the liquor, and the correlation between the factors and the year of liquor was small, so the first six factor can be used to predict the year of Chinese liquors. By selecting the factors with different contribution rates to predict the year of liquor, the average error can be reduced to 0.9 years.

辜姣, 陈国庆, 张笑河, 刘怀博, 马超群, 朱纯, 廖翠萃. 基于小波分解和因子分析的白酒香型和年份鉴定的研究[J]. 光谱学与光谱分析, 2018, 38(8): 2511. GU Jiao, CHEN Guo-qing, ZHANG Xiao-he, LIU Huai-bo, MA Chao-qun, ZHU Chun, LIAO Cui-cui. Classification and Year Prediction of Chinese Liquors Based on Wavelet Decomposition and Factor Analysis[J]. Spectroscopy and Spectral Analysis, 2018, 38(8): 2511.

本文已被 1 篇论文引用
被引统计数据来源于中国光学期刊网
引用该论文: TXT   |   EndNote

相关论文

加载中...

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!