摘要
以食用植物油为研究对象,建立了一种用近红外光谱技术鉴别食用植物油品种的方法。分别采用系统聚类法、主成分分析法、BP人工神经网络法进行了4个常用品种食用植物油的鉴别研究。结果表明,系统聚类法和主成分分析法均只能鉴别出4种植物油中的大豆油和花生油,不能识别玉米油和芝麻油,鉴别率仅为31.2%;利用BP人工神经网络法将60个校正集样品的11个主成分数据作为BP网络输入变量,建立的3层BP人工神经网络鉴别模型对4种植物油品种的鉴别效果最优,鉴别率为100%,表明BP人工神经网络法具有很好的分类和鉴别常用食用植物油品种的效果。
A new method for the discrimination of varieties of edible vegetable oil by means of near infrared spectroscopy（NIRS） was developed.Using hierarchical clustering method,principal component analysis,BP artificial neural network method（ANN-BP） to identify the four different varieties of edible vegetable oil.The results showed that the hierarchical clustering method and principal component analysis could only identify soybean oil and peanut oil,their recognition rates just reached 31.2%.With ANN-BP,the 11 principal components data of 60 calibrated samples were used as the inputs of ANN-BP,then the three layers ANN-BP discrimination model was built,and it was the best method to identify the four different varieties of vegetable oil,and the recognition rate for the four different varieties of edible vegetable oil was 100%.So it was reliable and practicable to use ANN-BP to identify the different varieties of edible vegetable oil effectively.
出处
《湖北农业科学》
北大核心
2011年第16期3383-3385,共3页
Hubei Agricultural Sciences
基金
华中农业大学科研专项(52204-02008)
关键词
食用植物油
近红外光谱
系统聚类法
主成分分析法
BP人工神经网络法
edible vegetable oil
near-infrared spectroscopy
hierarchical clustering method
principal component analysis
BP artificial neural network
