基于PLS-DA判别法对黑龙江大米产地溯源的研究

Study on Heilongjiang Rice Origin Traceability Based on Partial Least Squares Discriminant Analysis

为了开发更方便、准确、快速的大米产地溯源方法,更好地维护市场秩序,保护消费者的合法权益,研究运用近红外光谱技术并结合PLS-DA法(Partial least squares discriminant analysis,偏最小二乘判别分析)进行建模,对2012年黑龙江5个水稻主产区(五常、佳木斯、齐齐哈尔、双鸭山和牡丹江)的118份大米粉末样品进行近红外光谱的扫描,确定光谱预处理方法为9点二阶求导结合7点平滑,建模波长为全波长。主成分分析提取3个有效主成分。用预测样本集进行模型的验证,5个地域的预测正确率分别为87.5%,87.5%,100%,100%和100%。初步认定可用于2012年黑龙江大米产地溯源模型的建立。

It was in order to better maintain the market order and protect the legitimate fights and interests of consumersto develop more convenient, accurate and fast method for tracing the origin of rice, the study used partial least squares diseriminant analysis to build the model and combined with near infrared spectroscopy, 118 rice powder samples of different varieties from five areas （Wuehang, Jiamusi, Qiqihaer, Shuangyashan and Mudanjiang） scanned the near infrared spectroscopy from Heilongjiang in 2012 rice. Pretreatment method was 9 second order derivative with 7 smooth. The modeling wavelength was the whole model. The 3 main effective principal components extracted through principal component analysis to extract the 3 main effective component were used to establish model. The forecast samples were used toverified the accuracy of the model. The verification results of five areas were 87.5%, 87.5%, 100%, 100% and 100%. PLS-DA method could be used in Heilongjiang rice origin traceability model in 2012, preliminary.