利用自助拉丁配分和模糊规则对橄榄油进行分类研究

利用Nicolet 6700傅里叶变换近红外光谱仪测定了20个橄榄油样品的光谱数据。采用多元散射校正、Savitz-ky-Golay平滑以及二阶求导对原始光谱数据进行预处理以提高信噪比、消除背景和基线的影响。在(4500～12000)cm~(-1)波段范围内计算20个近红外光谱的标准偏差,以(5200～5300)cm~(-1)和(5600～6000)cm~(-1)作为输入建立模糊专家系统。建模过程中采用n次自助拉丁配分法进行评价分类和建模的结果,将更具统计意义的n次均值作为预测结果。利用模糊专家系统的分类树和定量预测都能成功地将20个橄榄油分为特级初榨和普通橄榄油2大类。结果表明,本文为橄榄油品质的鉴定提供了1种简便、快捷、准确的方法。

The classification of olive oil based on near infrared spectra by using boot-strapped latin partition and fuzzy theory

The NIR spectra of 20 olive oil samples are measured by using Nicolet6700 FTIR spectrometer. In order to improve the signal-noise ratio and remove the influences of background and baseline, the original spectra were pretreated by multiplicative scatter correction( MSC), Savitzky-Golay smoothing and the second derivative. The standard deviation of 20 olive oil NIR spectra, in the region of (120 00 ～4 500)cm~(-1), is computed. The data of NIR spectra in the region of (5 200 ～5 300)cm~(-1) and (5 600～6 000) cm~(-1) are employed as the input to build the fuzzy rale-building expert system(FuRES). Using n times bootstrapped Latin-partition to evaluate the results of classification and calibration models. The average values of n bootstrapped are meaningful results with statistical. The 20 olive oil samples are successful classified into extra olive and ordinary olive by using the classification trees and quantitative perdition model of FuRES. The results indicated that the method studied in this paper is a simple, fast and accurate approach for the discrimination of olive oil with different quality.