基于近红外光谱的橄榄油品质鉴别方法研究

发布时间：2018-03-13 12:17

本文选题：近红外光谱　切入点：橄榄油　出处：《光谱学与光谱分析》2016年09期 　论文类型：期刊论文

【摘要】：目前市面上销售的橄榄油主要分为特级初榨橄榄油和普通初榨橄榄油两类,为了鉴别两种不同品质的橄榄油,提出了一种应用siPLS-IRIV-PCA算法的橄榄油品质鉴别的新方法。基于橄榄油的近红外光谱数据,应用联合区间偏最小二乘法(siPLS)对橄榄油的近红外光谱进行了波长区间优选,使用交叉验证均方根误差(RMSECV)评估模型的性能并选择最优波长区间,通过迭代保留信息变量(IRIV)算法从最优波长区间中选择特征波长,根据选择的特征波长构建主成分分析(PCA)模型。对90组特级初榨橄榄油和90组普通橄榄油样本进行了判别鉴定。PCA将1 427个波长变量作为输入变量,前两个主成分贡献率为51.891 8%和26.473 2%;siPLS-PCA将408个波长变量作为输入变量,前两个主成分贡献率为56.039 1%和36.2355%;siPLS-IRIV-PCA将6个波长变量作为输入变量,前两个主成分贡献率为66.347 6%和32.304 3%。结果表明,与PCA和siPLS-PCA鉴别方法相比,siPLS-IRIV-PCA具有最佳的鉴别性能。
[Abstract]:At present, the olive oil on the market is mainly divided into two categories: extra virgin olive oil and ordinary virgin olive oil. In order to identify two different types of olive oil, A new method of olive oil quality identification based on siPLS-IRIV-PCA algorithm is proposed. Based on the near infrared spectral data of olive oil, the wavelength range of olive oil is optimized by using the combined interval partial least square method. The root-mean-square error (RMSECV) is used to evaluate the performance of the model and to select the optimal wavelength range. The characteristic wavelength is selected from the optimal wavelength range by iterative preserving information variable (IRIV) algorithm. The principal component analysis (PCAA) model was constructed according to the selected characteristic wavelengths. A total of 1,427 wavelength variables were used as input variables in 90 groups of virgin olive oil and 90 groups of common olive oil samples. The first two principal component contribution rates were 51.891 8% and 26.473,22-PCA with 408 wavelength variables as input variables. The first two principal component contributions were 56.039 1% and 36.235555%. The first two principal component contribution rates were 56.039 1% and 36.235555%. The first two principal component contribution rates were 66.347 6% and 32.3043 respectively. Compared with the PCA and siPLS-PCA discriminant methods, the siPLS-IRIV-PCA has the best discriminant performance.
【作者单位】：江苏大学电气信息工程学院;
【基金】：国家自然科学基金项目(31271875) 江苏省自然科学青年基金项目(BK20140538) 中国博士后科学基金项目(2014M550273)资助
【分类号】：O657.33;TS225.19

【相似文献】