摘要
寻找油茶籽油有关指标的快速检测方法,本研究应用红外光谱分析技术,结合人工神经方法,预测不同氧化程度油茶籽油的酸价、过氧化值和碘值。油茶籽油红外光谱经SG方法平滑,一阶导数等预处理。光谱数据和氧化指标实测值作为神经网络输入和输出,建立油茶籽油有关指标的定量模型。结果表明,酸价、过氧化值和碘值模型的相关系数(R)分别为0.914 2、0.964 9和0.964 2;均方根差(RMSE)分别为0.196 9 mg KOH/g、0.392 6 mmol/kg和1.010 5 g I/100 g。对模型进行仿真应用,酸价、过氧化值和碘值模型预测结果的相关系数(R)分别为0.981 9、0.993 0和0.983 3,预测标准偏差(RMSEP)分别为0.203 0 mg KOH/g、0.241 8 mmol/kg和0.602 8 g I/100 g。表明建立的模型可靠,预测效果好,能满足油茶籽油有关指标的快速检测要求。
A rapid method of determination on relevant indexes of camellia oil was developing. With Artificial Neural Network （ANN） models, infrared spectra （IR） were used to correlate with the camellia oil acid value, peroxide value and iodine value. The spectra data of camellia oil were preprocessed by the method of Savitzky - Golay （SG） and first derivative（FD）. Spectral data and relevant indexes data, as neural network inputs and outputs, were adopted to establish quantitative models of relevant indexes of camellia oil. The results showed that the correlation coefficient （R） of the model of acid value, peroxide value and iodine value were 0. 914 2, 0.964 9 and 0.964 2 respectively, the standard error （RMSE） of them were 0. 196 9 mgKOH/g, 0. 392 6 mmol/kg and 0. 392 6 gI2/100g respectively. After Simulation and application of the models of acid value, peroxide value and iodine value, the results show that the correlation coefficient of them were 0. 981 9, 0. 993 0 and 0. 981 9 respectively, the predict standard deviation （RMSEP） of them were 0.203 0 mgKOH/g, 0. 241 8/kg and 0.602 8 gI2/100 g respectively. Results suggested that the models were reliable with good accuracy and could meet the requirement of rapid determination of relevant indexes of camellia oil.
作者
王泽富
吴雪辉
江盛宇
章文
蓝梧涛
Wang Zefu1, Wu Xuehui1,2, Jiang Shengyu1, Zhang Wen1, Lan Wutao1(1.College of Food Science, South China Agricultural University, Guangzhou 510642;2. Guangdong Engineering Research Center for Oil - Tea Camellia, Guangzhou 51064)
出处
《中国粮油学报》
EI
CAS
CSCD
北大核心
2018年第3期119-125,共7页
Journal of the Chinese Cereals and Oils Association
基金
林业公益性行业科研专项经费项目(201504703)
关键词
油茶籽油
人工神经网络
酸价
过氧化值
碘值
红外光谱
camellia oil, artificial neural network, acid value, iodine value, peroxide value, infrared spectroscopy
