| With the people's health awareness, food security problem was given more and more attention. Milk and dairy products has become the closely related main food to people, are an important food product related to the healthy growth of people especially the babies, therefore the quality of milk and dairy products have become major community concern. Emerged in 2008 cow dairy melamine adulteration case, make people are particularly concerned about food safety issues. Fat and protein content of the decision are the core indicators of milk quality, but also affect the measurement accuracy of the key aspects, the vast majority testing work of milk quality is concentrated in these two indicators. Therefore, how to precisely detect all kinds of milk ingredients related to the people's lives and health is an important question.Because of the complexity of the mile detection, the detection equipment at home and abroad are almost the same, the detection accuracy of milk need to be further improve and research. At present, front-end equipment in milk testing has become more sophisticated. Because of the infection of front-end detect hardware astrict and and temperature, humidity, etc. All kinds of testing equipment in the process of the detection accuracy exist uncertainty and error.Therefore, how to eliminate such errors, improve accuracy of detection has become a subject to be studied.BP network because of its good nonlinear approximation ability and generalization ability as well as easily adaptable are widely used. For milk testing, hardware inherent drawbacks, this paper, we introduce BP neural network to improve the measurement accuracy of instruments. For those data that have been extracted from the back-end of the equipment, we using BP neural network to training the non-linear model.Thesis give the BP algorithm software implementation and the Matlab simulation testing. Through theoretical analysis and experiments proved that the design of BP network is basically feasible in this article, compared to the traditional milk detection technology, improved protein and fat detection accuracy. |
