Maitai Origin Recognition Technology Research

The distilled liquor samples were treated by Microwave digested pre-treatment. Samples from different geographic origin and corporations were analyzed by ICP-OES and ICP-MS, stable isotope ratios of certain elements and the various elements in the distilled liquors actually provide complementary information about the provenance which shows their origin, the DATA acquired can be combined with multivariate statistics. It is concluded that at the multi-elements aspect, there is distinguish difference among the distilled liquors form different geographic origin or provenance. For example, the Boron isotope ratio reflects some kind of geographic characteristic. As the multi-dimension, complexity and non-liner feature of the information get from the distilled liquor, some kind of system clustering methods appear unstable consequences of subjective defects. To avoid the disadvantage, utilizing the capacity of Back Propagation Neural Network, self-organizing, self-studying, distribute saving, fault-tolerant, robustness and nonlinear function approximating, programming were implemented by M language, proposed a 3-layer back propagation neural network. After constructing, training, stimulating test and validation, the BP network is able to discriminate samples from different provenance with multi-elements input vector matrix. Simple, not time consuming and convenience, BP network can be a strong engineering tool for authenticity and geographic origin discrimination.