Method for discriminating adulterated raw and fresh milk by using near infrared spectrum

Abstract

The invention provides a method for discriminating adulterated raw and fresh milk by using near infrared spectrum, which comprises the following steps of: (1) collecting the near infrared spectra of various raw and fresh milk samples and adulterated milk samples, and correspondingly establishing a raw and fresh milk spectrum library and an adulterated milk spectrum library; (2) establishing a discrimination model by using the near infrared spectra in the raw and fresh milk spectrum library and the adulterated milk spectrum library, and correcting the discrimination model so as to obtain an optimal discrimination model; (3) collecting the near infrared spectrum of an unknown milk sample which is adulterated or not adulterated, and discriminating the near infrared spectrum of the unknown milk sample by using the optimal discrimination model so as to obtain the discrimination result of the unknown milk sample, wherein the discrimination model is established by using the SVM (Support Vector Machine) method in the step (2). With the method, the discrimination accuracy on adulterated milk containing various unknown adulterants is greatly increased, and the optimal discrimination model is updated and maintained by expanding and updating the near infrared spectra in the raw and fresh milk spectrum library and the adulterated milk spectrum library.

Description

technical field [0001] The present invention relates to a method for detecting raw and fresh milk, and more specifically, relates to a method for identifying the adulteration of fresh and fresh milk by using near-infrared spectroscopy. Background technique [0002] Adding water, dextrin, non-dairy creamer, melamine and other low-cost substances to fresh milk for commercial profit is a common illegal act. [0003] Existing test strips and chromatographic detection methods are usually used for separation and analysis of a specific adulterant. Therefore, it is time-consuming and labor-intensive to identify different adulterants after different pre-processing steps. [0004] Recently, some scholars have used commercial milk to carry out some studies on the qualitative identification or quantitative analysis of adulterated milk using near-infrared technology and achieved good results. However, commercial milk has undergone homogenization, sterilization and other processes and ad...

AI Technical Summary

Problems solved by technology

[0005] The inventor of this patent once collected 162 unprocessed raw milk from various pastures in Shanghai and surrounding areas during June-July 2010, and prepared the same number of melamine-containing, For adulterated milk solutions of urea and ammonium nitrate, the near-infrared spectra of fresh milk samples and adulterated milk samples were collected respectively. By comparing different spectral preprocessing methods and modeling methods, the best near-infrared discrimination model was established. The results showed that the The discriminant model established by near-infrared commercial software has a correct rate of about 40% for unknown milk samples, and the model parameters cannot be updated. The self-developed IS-KNN method (improved and simplified K-nearest neighbor method) was used to establish The discriminant model can have a maximum accuracy of 100% for adulterated milk with known adulterants, but the correct rate for adulterated milk with multiple unknown adulterants is usually 80% to 90%. , there is still room for improvement in the existing methods for the correct rate of discrimination of adulterated milk containing a variety of unknown adulterants and the update and maintenance of the parameters of the discriminant model

Images