基于Fisher判别法对黑龙江大米产地溯源

Study on the origins of Heilongjiang rice based on Fisher discriminate method

为了探索大米产地鉴别可行性,维护市场秩序和消费者的合法权益,该研究采用Fisher判别法(Fisher discriminant method,费希尔判别法)进行建模并结合近红外光谱技术,对2015年黑龙江5个水稻主产区(五常、佳木斯、齐齐哈尔、双鸭山、牡丹江)的118份大米粉末样品进行近红外光谱的扫描,光谱预处理方法为9点二阶求导结合5点平滑,建模波长为全波长。对模型采用留一交叉验证和预测样本集进行验证,5个地域的验证结果分别为94.4%、94.4%、91.7%、91.7%、94.4%和87.5%、87.5%、87.5%、100%、100%。预测结果达到80%以上,初步认定近红外光谱指纹分析技术可用于黑龙江大米产地溯源。

In order to explore the feasibility of identification of the origins of Heiloingjiang rice to protect consum- er＇s rights, food safety issue and legitimate issues, the study used Fisher discrimination method combined with near infrared spectroscopy to build a model. There were 118 rice powder samples from different varieties in five areas （Wuchang,Jiamusi, Qiqihaer, Shuangyashan） were scanned by near infrared spectroscopy in 2015. Pretreatment method of spectroscopy data is based on ＂9-point derivative combined with 5-point smoothing＂. The wavelength is the whole spectrum. The cross validation and the predicting samples were used to verified the accuracy of the model. The verification results of five areas were 94.4%, 94.4%, 91.7%, 91.7%, 94.4% and 87.5%, 87.5%, 87.5%, 100% , 100%. Predicting accuracy was more than 80% , and Fisher discrimination method was primarily assigned to use in heilongjiang rice origin trace identification.